Skip to content

feat(mcp): session_set_project sets the project of a session the caller created - #15443

Closed
Pearcekieser wants to merge 1 commit into
kirodotdev:mainfrom
Pearcekieser:feat/mcp-session-set-project
Closed

Pearcekieser wants to merge 1 commit into
kirodotdev:mainfrom
Pearcekieser:feat/mcp-session-set-project

Conversation

@Pearcekieser

@Pearcekieser Pearcekieser commented Sep 30, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: an agent can set the project directory of a session it created, for example to point a worker at a new git worktree.

set_project only changes the calling session's own project. A conductor that opens a worker with session_create cannot move that worker to a new worktree afterwards. It has to ask the user to change the project in the worker's tab, or tell the worker to call set_project on itself, which costs a turn and only works once the worker is already running.

Why it matters

Worktree-per-item is how conductors keep parallel workers from editing the same checkout. Without this, the conductor either needs a person to click through each worker tab, or has to spend a worker turn on setup before the real work starts.

Not a goal

  • Changing the project of a session the caller did not create, including the person's own sessions.
  • New path rules or an allowlist of project roots. The path gets the checks set_project already applies.
  • Changing the target's workspace, agent or model.

What changed (motivation → approach → change)

A project change restarts the target's session in a new CWD, so the new verb reuses set_project's own write and is fenced harder than the other session verbs.

  • session_set_project(target, path) is a Tool(...) row in _session_tools() in src/kiro_crew/mcp_dashboard.py (identity strict, one declared route). It calls POST /api/session-control/set-project, which is registered in src/kiro_crew/dashboard/server_runtime/mcp_routes.py and added to _STRICT_INTERNAL_API_PATHS in src/kiro_crew/dashboard/server.py. The tool description and its result both say the target resets at its next turn boundary.
  • set_project_target in src/kiro_crew/dashboard/session_control.py runs authorize_target, then refuses a session the caller did not create (not_creator, checked whatever the fence verdict or the global switch say), a pinned one (pinned_target), a read-only relay archive (relay_archive_read_only, the shared refusal from feat(remote-crew): delete the turn relay; old relay chats are read-only #17662), and a busy one (target_busy: a turn in flight, attached sub-agents, queued messages, or a switch handler holding the slot lock). A closed session does not resolve (target_not_found).
  • The path is resolved only after those gates, so a refused caller cannot probe the filesystem. Resolving awaits, so the gate, the refusals and the idle probe run again synchronously right before the write, under the same metadata lock the pin and folder routes hold. The attached sub-agent probe runs again as the first step inside that lock, because its earlier answer can go stale while the lock is awaited.
  • A change is persisted with a forced save before the call returns, because an idle target may not save again before a gateway restart. A refused save (the session was deleted or rebound mid-persist) rolls the project and the reset back and answers session_gone; a failed save (lock or I/O error) rolls back and answers persist_failed. The rollback runs only while the project and the reset it armed are both still this call's values. If the target consumed the reset during the save, it already runs in the new directory, so the project stays, the slot is marked dirty for the periodic flush, and the error says so.
  • Every exit after the commit leaves a terminal SEL record. A CancelledError (gateway shutdown, client disconnect) raised by the forced save or the recent-projects write skips except Exception and _audit_denials, so both awaits catch it, record the decision as allowed with interrupted set to the exception name, mark the slot dirty so the flush finishes the save, and re-raise.
  • _set_project in src/kiro_crew/dashboard/session_directive_apply.py is split into resolve_project_path (sensitive path before and after realpath, must be a directory, data-home overlap) and commit_project (write, and arm the deferred reset on a change). The directive and the new verb both call them; the directive's results and audit outcomes are unchanged.
  • The route validates field types in src/kiro_crew/dashboard/handlers/session_control.py; the MCP schema in src/kiro_crew/validation.py reuses set_project's absolute-path shape, with path required. There is no clear: nothing needs a conductor to unset a worker's project.
  • The tool is on the channel-agent block list (src/kiro_crew/channel.py), recorded as withheld from conductor auto-approval (src/kiro_crew/agent.py), titled in src/kiro_crew/data/mcp_tool_titles.json, and documented in src/kiro_crew/docs/session-control.md, src/kiro_crew/docs/index.md, docs/architecture/mcp.md, docs/system-specs/modules/session-control.md and the Session control row of docs/feature-map/README.md.

Backwards compatibility

Compatible: a new tool and route. The set_project directive keeps the same checks, messages and audit outcomes; its body moved into two shared helpers.

Tests

test/test_session_control_set_project.py (new, 45 cases): a created idle target takes the realpath and arms the reset; the change is saved with force=True against the resolved history key before the call returns, an unchanged project is not saved, a refused or failed save rolls back without erasing a newer writer's project or a same-text write from the dashboard route (identity-token check), and a save that fails after the target consumed the reset keeps the new project; a cancel during either post-commit await still writes one allowed audit record and leaves the slot dirty, and a post-commit refusal is audited once as denied; the same project again arms nothing; sessions the caller did not create, pinned, closed, self, channel-linked and relay-archive targets are refused; a turn, queued messages, the slot lock and attached sub-agents refuse with target_busy; a turn starting, a pin or a channel link landing while the path resolves is caught by the second check, and a sub-agent attached while the lock is awaited is caught by the in-lock probe; sensitive paths (direct and through a symlink), a missing directory and a data-home overlap are refused; a refused caller's path is never resolved; an empty path is refused, and the MCP schema refuses a call without path; the route refuses wrongly typed fields and is on the strict internal set; the tool sends the verified key and says the target resets. Mutation check: removing the creator refusal fails the three scope tests.

Updated: the advertised-set pins in test/test_mcp_dashboard_registration.py and test/test_mcp_dashboard_folders.py, the session-control count in test/test_session_control_doc.py, the dashboard tool-call table in test/test_session_control.py, and the reservation-reader list in test/test_slot_running_readers.py, which now registers _set_project_busy as a busy helper so it covers set_project_target itself.

Manual verification

N/A. The change is gateway-side state handling behind an MCP tool with no UI; the unit tests drive the verb, the shared path check, the route and the tool text. After the port onto the Tool(...) table, 671 tests across the ten related files and 805 across the MCP dashboard, MCP tools, agent contract and docs suites pass locally.

Related Issues

Fixes #15440

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title (feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)
  • Existing tests pass and new tests added for new functionality
  • Self-review completed; code follows project style guidelines
  • Documentation updated (if applicable)
  • No secrets, credentials, or internal references in the diff

@github-actions github-actions Bot added fork Pull request from a fork (external contributor) readiness: action required A blocking check or review needs attention labels Sep 30, 2026
@dwu96

dwu96 commented Sep 30, 2026

Copy link
Copy Markdown
Contributor

👋 Hi! This PR is currently in draft status. Workflow runs won't be auto-approved until it's marked as ready for review.

When you're ready, click "Ready for review" and the workflows will be approved on the next cycle automatically.

1 similar comment
@dwu96

dwu96 commented Sep 30, 2026

Copy link
Copy Markdown
Contributor

👋 Hi! This PR is currently in draft status. Workflow runs won't be auto-approved until it's marked as ready for review.

When you're ready, click "Ready for review" and the workflows will be approved on the next cycle automatically.

@Pearcekieser
Pearcekieser force-pushed the feat/mcp-session-set-project branch from eec41e0 to 838818d Compare September 30, 2026 16:34
@Pearcekieser
Pearcekieser marked this pull request as ready for review September 30, 2026 16:34
@Pearcekieser
Pearcekieser requested a review from a team as a code owner September 30, 2026 16:34
@Pearcekieser

Copy link
Copy Markdown
Contributor Author

Intent: let an agent set the project directory of an idle, unpinned session it created, through the same path checks set_project applies to a session's own project, so a conductor can point a worker at a new worktree without a person clicking through its tab.
Not a goal: reaching sessions the caller did not create (including the person's own), new path rules or root allowlists, or changing the target's workspace, agent or model.

@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed readiness: action required A blocking check or review needs attention labels Sep 30, 2026
@github-actions

github-actions Bot commented Sep 30, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5, fork) — 🟡 CONCERNS

Design-level review of 5b0478181d17ba05ba7ce1de97e933dc525bda24 via the fork AI-review pipeline — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

Design-Verdict: CONCERNS

The PR never weighs an obvious simpler option: a project argument on session_create, which needs none of the busy, race and rollback handling.

Watch

  • Simpler alternative not considered. The goal is to "point a worker at a new git worktree after session_create". A brand-new session has no turn, queue or sub-agents to cut into. Setting the project afterwards creates a window that a turn or a pin can land in. That window is why set_project_target needs about 275 lines: a pre-lock and an in-lock sub-agent probe, a second authorize_target, a _CommitToken identity check, three rollback branches and CancelledError audit handling. The cost is a large new mutation path on another session's CWD, with failure states that are hard to reason about (for example "already started on the new project; the save failed and will be retried"). The description doesn't say why the per-item worktree can't be set when the worker is created, or why existing folder project_dir filing doesn't cover it.
    Clears when: the description names a conductor flow that must re-point an existing worker, or the verb becomes a project field on session_create that goes through resolve_project_path.

[DESIGN-REVIEWED] 5b04781

Description read

[DESCRIPTION-READ] 0a1f99701d62651a24cbb8d344abb993805a0000be30cf514db88720dbb72ef6

That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Sep 30, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5, fork) — 🟡 CONCERNS

Premise-level review of 5b0478181d17ba05ba7ce1de97e933dc525bda24 via the fork AI-review pipeline — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

First-Principles-Verdict: CONCERNS

The tool is pitched for conductors, yet the diff withholds it from them and no conductor skill calls it. A session_create birth-time project would avoid the reset machinery.

Not justified as shipped

  • Item 1 — inherited: the description argues "Worktree-per-item is how conductors keep parallel workers…", but the agent.py hunk says "no unattended conductor step needs it". The pattern session_set_project matches 0 files in src/kiro_crew/builtin_skills. No in-repo step moves a worker that is already running.
    Subtraction: replace the after-the-fact verb with the path set at creation. A session is idle when it is created, so this deletes _set_project_busy, the probe inside the lock, the rollback branches (three refusal messages), _audit_interrupted_commit, and the pinned_target/not_creator gates.
    Clears when: feat(mcp): session_set_project sets the project directory of a session the caller created #15440 or a conductor step names a case that needs a running worker moved, or the verb is replaced by a project set at birth.
  • Item 6 — undeclared: an agent-set worktree path now goes into the person's recent-projects list (save_recent_project(project)).

What this change ships

Inventory (7 items) — 5 justified

Intent: let an agent point a session it created at a different project directory, such as a new worktree. ADDITION.

  1. Agents can move a created session's project directory — inherited, consumer excluded, smaller alternative
  2. Target refused when busy, pinned, not created by the caller, or a relay archive — justified
  3. Target cold-starts in the new CWD at its next turn — justified
  4. Change force-saved before acknowledging, rolled back on failure — justified
  5. The set_project directive's body split into shared helpers, behaviour unchanged — justified
  6. Recent-projects list gains agent-chosen paths — undeclared
  7. Tool blocked for channel agents, withheld from conductor auto-approval — justified

[FIRST-PRINCIPLES-REVIEWED] 5b04781

Description read

[DESCRIPTION-READ] 5fbac07197a28f268af5f8842e91ca8d24dedd1f76f6fa09fd58574b2d166e8d

That sha256 names the PR description this verdict read. Recompute it with .github/scripts/pr-description-capture.sh; if it differs, the description changed after this verdict was formed and any finding drawn from the description is unproven.

@github-actions

github-actions Bot commented Sep 30, 2026 •

Copy link
Copy Markdown
Contributor

GPT 6.1 Review (fork) — ✅ no blocking findings

Reviewed 5b0478181d17ba05ba7ce1de97e933dc525bda24 via the fork AI-review pipeline; updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 5b04781

@github-actions

github-actions Bot commented Sep 30, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5.5 Review (fork) — ✅ no blocking findings

Reviewed 5b0478181d17ba05ba7ce1de97e933dc525bda24 via the fork AI-review pipeline; updated in place on each push.

Review details

No findings block this merge. One advisory finding.

FINDING — src/kiro_crew/dashboard/session_control.py:6170 — if a turn consumes the armed reset while the forced save is running and the save then fails or is refused, the change has already taken effect (the target runs in the new CWD and slot.project stays at the new value). But the if ours and not reset_unchanged: raise SessionControlError(...) branch is inside _audit_denials, so the SEL records it as outcome="denied", with no "allowed" record and no _push(state). The cancel path labels the same committed state "allowed" → Fix: in that branch, emit _audit(..., outcome="allowed", detail={"project": project, "changed": True, "persist": code}) and call _push(state) before raising, and raise outside the _audit_denials region (or skip it with a flag) so it is not recorded a second time as denied.

[OPUS-REVIEWED] 5b04781

@github-actions github-actions Bot added merge conflict Branch has merge conflicts with its base — author must resolve before merge readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Sep 30, 2026
@Pearcekieser
Pearcekieser force-pushed the feat/mcp-session-set-project branch from 838818d to 44883ce Compare September 30, 2026 18:01
@Pearcekieser

Copy link
Copy Markdown
Contributor Author

self-added: yes
mechanism: rollback guard now keyed on both the project and the reset this call armed

  • rollback can corrupt project state after reset consumption

fixed in 44883ce. set_project_target records the reset key commit_project armed and rolls back only while slot.project AND _pending_reset_history_key still hold this call's values.
If a turn consumed the reset during the save, the target already runs in the new CWD, so the project is kept, the slot is marked dirty for the periodic flush, and the error says the target started on the new project and the save will be retried.
Tests cover a refused save and an I/O failure after the reset was consumed, plus the existing rollback cases.

@Pearcekieser

Copy link
Copy Markdown
Contributor Author

self-added: yes
mechanism: removed the clear field from session_set_project

  • clear rides on symmetry with the set_project directive's shape

fixed in 44883ce by taking the subtraction: clear is gone from the tool schema, SESSION_SET_PROJECT_SCHEMA, the route, set_project_target and the docs, and path is required.
The self-directed set_project keeps its own clear; its schema is untouched.

@Pearcekieser

Copy link
Copy Markdown
Contributor Author

Rebased onto main (c1a02ce). main gained session_end_wait (#15166), which conflicted only in the tool lists in the docs; the resolution lists both tools and the counts are now 25 tools and 15 session-control tools. The same push carries the fixes for the two findings above. CI is re-running.

@github-actions github-actions Bot added readiness: checking Automated validation is still running readiness: action required A blocking check or review needs attention and removed readiness: action required A blocking check or review needs attention merge conflict Branch has merge conflicts with its base — author must resolve before merge readiness: checking Automated validation is still running labels Sep 30, 2026
@Pearcekieser
Pearcekieser force-pushed the feat/mcp-session-set-project branch from 44883ce to e0e6882 Compare September 30, 2026 20:30
@Pearcekieser

Copy link
Copy Markdown
Contributor Author

self-added: yes
mechanism: attached sub-agent probe re-run as the first step inside the metadata lock

  • Sub-agent verdict goes stale while awaiting the metadata lock

fixed in e0e6882. set_project_target now awaits subagents_attached_async again as the first statement inside _slot_meta_txn_lock, then runs the synchronous re-checks (gate, replacement, creator/pinned refusal, idle probe) and commit_project with no await in between.
The pre-lock probe stays as a fast refusal. A new test makes the in-lock probe report an attached child after the first probe passed and asserts target_busy with no project change and no armed reset.

@Pearcekieser

Copy link
Copy Markdown
Contributor Author

self-added: yes
mechanism: _set_project_busy registered as a busy helper in the reservation-reader ratchet

  • busy-reader ratchet does not cover set_project_target

fixed in e0e6882. _set_project_busy is now in _BUSY_HELPERS as a reservation reader and set_project_target is registered in the expected set, keeping the helper's own entry.
Mutation check: deleting both _set_project_busy refusals from set_project_target makes test_running_and_turn_running_slot_readers_are_enumerated fail.

@github-actions github-actions Bot added readiness: checking Automated validation is still running readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Oct 7, 2026
@Pearcekieser
Pearcekieser force-pushed the feat/mcp-session-set-project branch from d00a814 to d6a5c51 Compare October 7, 2026 02:45
@Pearcekieser

Copy link
Copy Markdown
Contributor Author

self-added: yes
mechanism: the save-failure rollback in set_project_target compared slot.project == project by value, so a same-path write by api_chat_slot_project during the save could not be told apart from this call's commit

  • rollback erases a newer same-path project selection

fixed in d6a5c51. set_project_target now commits a per-call _CommitToken(project) through commit_project and rolls back only when slot.project is committed_project, the identity check api_chat_slot_project already uses. When another writer has set the same path, the project stays, the slot is left dirty for the periodic flush, and the caller is told the target is already on the new project and the save will be retried. Opus 5.5 raised the same defect as an advisory on this head; this one fix covers both.
Test that fails first: test_a_refused_save_keeps_a_same_text_write_from_another_writer writes an equal-text, distinct string during the save and refuses it. On d00a814 the rollback restored /before; it passes now.
This span id is on its fourth round on this file, each a different defect in the same post-commit rollback region. The First Principles review proposes removing that region by setting the project at create time; see the reply to it.

@Pearcekieser

Copy link
Copy Markdown
Contributor Author

self-added: yes
mechanism: the save-failure rollback compared slot.project by value, so it could not tell a same-text write from api_chat_slot_project apart from this call's commit

  • same-text write erased by the value-compare rollback

fixed in d6a5c51, with the same change as the GPT 6.1 blocking finding on this line: the commit is a per-call _CommitToken(project) and the rollback requires slot.project is committed_project. test_a_refused_save_keeps_a_same_text_write_from_another_writer fails on d00a814 and passes now.

@github-actions github-actions Bot added readiness: checking Automated validation is still running readiness: action required A blocking check or review needs attention merge conflict Branch has merge conflicts with its base — author must resolve before merge and removed readiness: action required A blocking check or review needs attention readiness: checking Automated validation is still running labels Oct 7, 2026
@Pearcekieser
Pearcekieser force-pushed the feat/mcp-session-set-project branch from d6a5c51 to f544216 Compare October 7, 2026 10:15
@Pearcekieser

Copy link
Copy Markdown
Contributor Author

Rebased onto main f6d32b5b57 with no behaviour change. New head: f544216a54.

The conflict came from #15158 (chat_folder_update), which edited the same dashboard tool lists in docs/architecture/mcp.md and src/kiro_crew/docs/index.md. Both entries are kept, and the tool count in index.md and session-control.md is now 31. 781 targeted tests pass locally.

The open GPT finding (span 36bec42fc986) on the save-failure rollback is unchanged; that part waits on a design decision.

@github-actions github-actions Bot removed the merge conflict Branch has merge conflicts with its base — author must resolve before merge label Oct 7, 2026
@github-actions github-actions Bot added the merge conflict Branch has merge conflicts with its base — author must resolve before merge label Oct 7, 2026
@Pearcekieser
Pearcekieser force-pushed the feat/mcp-session-set-project branch from f544216 to 8cab878 Compare October 8, 2026 01:15
@Pearcekieser

Copy link
Copy Markdown
Contributor Author

Rebased onto current main (db083f5d3d), no behaviour change. The text conflicts were the Session control row in docs/feature-map/README.md (both sides' edits kept) and the MCP route-table and boot-trace pins in test/test_dashboard_server_composition_contract.py: main's new routes plus this PR's one set-project route give 243 rows (digest re-pinned) and one more _deferred in _DASHBOARD_BOOT and _API_BOOT. #17662 removed _ChatSlot.is_remote, so set_project_target now refuses a relay archive through the shared is_relay_archive check (relay_archive_read_only, 409), the same refusal session_set_model and session_reload use, and the unraised remote_target_unsupported row is gone from the refusal table. The 781 targeted tests pass. The PR is still parked on the rollback design decision.

@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed merge conflict Branch has merge conflicts with its base — author must resolve before merge readiness: action required A blocking check or review needs attention labels Oct 8, 2026
@github-actions

github-actions Bot commented Oct 8, 2026 •

Copy link
Copy Markdown
Contributor

UX Review (Opus 5.5, fork) — ⏭️ skipped

Revision 5b0478181d17ba05ba7ce1de97e933dc525bda24 touches no user-facing surface (no changes under website/ or committed screenshots), so the UX review was skipped. Advisory — does not block merge.

@github-actions github-actions Bot removed the readiness: checking Automated validation is still running label Oct 8, 2026
@iamwhatever

Copy link
Copy Markdown
Contributor

This PR adds session_set_project, which lets a conductor move a worker it created to a new project directory (for example a fresh git worktree). It reuses set_project's existing write and path checks behind a creator-only fence. I read the description, the file list and the lane verdicts on head 8cab878. The change goes straight at the gap, so I have no root-cause objection. It is on hold because GitHub reports mergeable=CONFLICTING (DIRTY, label "merge conflict"). It is also a permission-sensitive verb: one agent changes the working directory of another session and restarts it there, and the Design and First Principles lanes both say CONCERNS on the current head. GPT and Opus report no blocking findings. Please rebase onto main, answer the Design and First Principles items for the new head, and have a maintainer review the ownership fence, since this widens what one agent can do to another session. — crew-pr-reviewer

@Pearcekieser

Copy link
Copy Markdown
Contributor Author

Thanks for the read. Status on each item:

The rebase is done locally: one commit on 51371a0d8d, no behaviour change, 781 targeted tests and the black gate pass. The only text conflict was the feature-map Session control row. It's waiting for a free slot in our fork's CI queue and goes out at the next quarter-hour after one frees.

The Design and First Principles CONCERNS both point at the same thing: the tool is withheld from unattended conductors, and a project argument on session_create would cover worktree-per-item without the post-create reset and rollback logic. That is a scope decision for the PR owner, and it is open together with a known race in the save-failure rollback (GPT span 36bec42fc986, cold-start turn captures the new CWD while the forced save is pending). I won't change the rollback until that call is made, and I'll answer both lanes against whichever head carries it.

Agreed on a maintainer review of the creator-only fence once the design is settled.

@Pearcekieser

Copy link
Copy Markdown
Contributor Author

Rebased onto current main again (aee834cc85), no behaviour change. Main had moved after readiness passed on 8cab8789d6; the one text conflict was docs/feature-map/README.md, where main edited only the Work ledger row and this PR only the Session control row, so I kept main's Work ledger row and this PR's Session control row. The 781 targeted tests and the black gate pass on the new head. The PR is still parked on the rollback design decision.

@Pearcekieser

Copy link
Copy Markdown
Contributor Author

Rebased onto current main (96c00c10cc), no behaviour change. The one text conflict was the Session control row in docs/feature-map/README.md: main added reasoning-effort wording to the set-model entry and this PR adds the set-project entry and its endpoint, so both edits are kept. The 782 targeted tests and the black gate pass on the new head. The PR is still parked on the rollback design decision.

…er created

Adds session_set_project to kirocrew-dashboard and POST /api/session-control/set-project. The verb reuses the set_project directive's path check and commit (split into resolve_project_path and commit_project), and reaches only unpinned, idle sessions the caller created. The target's conversation resets at its next turn boundary.
@Pearcekieser

Copy link
Copy Markdown
Contributor Author

Rebased onto current main (87c429aa69), no behaviour change. Main added session_retry, which edited the same tool lists and docs as this PR. Both tools are kept everywhere: the kirocrew-dashboard list in docs/architecture/mcp.md, the Session control row in docs/feature-map/README.md (word-level merge), and both the tool list and a session_retry section followed by the session_set_project section in src/kiro_crew/docs/session-control.md. The counts move to 32 dashboard tools and 17 session-control tools. In test/test_dashboard_server_composition_contract.py main's 243 routes plus this PR's set-project route give 244 rows, the digest is re-pinned, and each boot list gets one more _deferred. With the set-project row removed the table hashes to main's own digest. The 782 targeted tests, the black gate and the feature-map gate pass on the new head. The PR is still parked on the rollback design decision.

@iamwhatever

Copy link
Copy Markdown
Contributor

Thank you, @Pearcekieser, for this and for keeping it rebased so diligently. We are closing it for now. The need is real, but the Design and First Principles reviews pointed to a smaller shape: a project argument on session_create, which avoids restarting another session and the rollback logic. #15440 is parked while that choice is made. The need stays tracked in #15440, and a PR for the smaller shape would be welcome.

The branch is left untouched, so you can reopen this PR at any time, or open a fresh one on current main, if you want to pick it back up.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

fork Pull request from a fork (external contributor) merge conflict Branch has merge conflicts with its base — author must resolve before merge

Projects

None yet

Development

Successfully merging this pull request may close these issues.

feat(mcp): session_set_project sets the project directory of a session the caller created

5 participants