io.github.entire-vc/evc-mesh-mcp

Add a comment to a task. If the body @-mentions someone, the response carries a `delivery` array — one entry per mentioned handle — reporting whether it actually reached a path they consume (their task queue, a notification) or was skipped/failed and why; a `hint` field suggests the fix when there is one (e.g. assign the task). Omitted entirely when the comment mentions nobody.

Add a dependency between two tasks.

Link a task to a pull request, commit, or branch. This is what makes the task↔PR join real: a task with no VCS link cannot be matched to the code that implements it, so PR-driven status automation and any 'what shipped for this task?' report simply will not see it. Call it as soon as the PR exists. Only task_id and url are needed — provider, link_type and external_id are inferred from a GitHub or GitLab URL. If the PR is ALREADY merged (or closed) by the time you call this — e.g. you finished, merged, and are linking retroactively — pass status='merged' (or 'closed'). Without it the link starts as 'open' and the done-evidence gate will block move→done on it forever: no GitHub webhook fires for a merge that happened before the link existed.

Assign a task to a user or agent.

Atomically acquire an exclusive lock on a task. Prevents other agents from checking out the same task simultaneously. The lock is TTL-based and will expire automatically after ttl_minutes (default 120). Use before starting work on a task to ensure exclusive access.

Release a human gate. Read human_gate_info on get_task first and go by clear_path. clear_path="clear_endpoint" means YOU armed this gate through set_human_gate and it carries no marker comment — this tool releases it, and a withdrawal comment would be a silent no-op. clear_path="withdraw_marker" means the ask lives in a "Blocking @" comment: this tool refuses, and you take it down by posting a short negator comment instead. Everything else is user-only — a gate a human armed, or one raw-armed via PATCH/UI with no author — and an agent key gets a 403 naming the exit it CAN reach: record the human's answer via a human-gate decision. Re-read human_gate after any release; a posted comment is not a cleared gate.

Comment on a document. To comment on a specific passage, pass quote with the text exactly as the document reads it — the server finds it and anchors the comment there, so you never compute a position yourself (there is no offset parameter, and a position you calculated would silently point at the wrong sentence). Without quote the comment is on the whole document. Your comment appears in the same thread humans see in the document UI.

Create a document in a project. Returns its metadata and version — the version is what update_doc takes as base_version, so a create followed by an edit needs no read in between. The body you sent is not echoed back.

Creates a recurring task schedule that automatically spawns task instances on a schedule. Each instance gets access to the previous instance's summary. Use this for regular automated work: weekly reports, daily checks, periodic audits.

Create a subtask under a parent task. Set status_slug for initial status (defaults to the project's default status, NOT the parent's status).

Create a new task. Check get_my_tasks and list_tasks FIRST to avoid duplicates. Set status_slug for initial status (defaults to project's first status).

Delete a recurring task schedule. Existing task instances are not affected.

Export the current workspace configuration as YAML, including rules, project templates, and settings.

Push the expiry of an existing checkout_task lock forward, for work that runs longer than the original ttl_minutes. Requires an active checkout in this session (the cached checkout_token from checkout_task) — fails if the lock was never acquired here, already released, or already expired. Server clamps ttl_minutes to [1, 240].

Delete a memory entry. Agents can only delete their own agent-scope memories.

Get artifact details. The bytes are never inlined: download them with the two GETs below. Downloading an artifact is two GETs. Step 1: GET <base>/api/v1/artifacts/<id>/download with header X-Agent-Key: <your agent key> -> 200 JSON {"url": "<presigned URL>"}. Step 2: GET that url with NO headers -> 200, the file bytes. Pitfalls: on step 1 only X-Agent-Key is accepted (X-API-Key and Authorization: Bearer give 401); on step 2 any extra header, Authorization in particular, breaks the presigned signature (400). The artifact's download_path is step 1's path. Never fetch browser_only_url with an agent key: it is a human page and answers 401 by design.

Get effective assignment rules for a project, merged from workspace and project level with source annotations.

Query the canonical knowledge layer: returns curated facts, decisions, and strategy docs for a topic, merged from project_memories (key canonical:*) and workspace_memories (kind:canonical). Excludes ephemeral session-checkpoints. Slug aliases are resolved automatically (e.g. mesh-dev == evc-mesh). Call before authoring any doc that might conflict with existing canonical knowledge.

Fetch canonical decisions broadcast since a given time. Call at ACP step 6 (session start) to catch up on owner directives since your previous session. Returns only privacy:public records targeted at you or all agents.

Get RECENT ACTIVITY for a project (last 24h by default): event stream with summaries, decisions, errors, plus accumulated project knowledge. Use for ACP Step 4 — what happened recently. For searching specific knowledge, use recall.

Read a document. By DEFAULT returns metadata plus the outline (headings) and NOT the body — a document is far larger than a task, and a body you read stays in your context for the rest of the session. Read the outline first, then pass section="<heading>" for just that part; body=true returns the whole page and should be the exception. The returned version is what update_doc takes as base_version.

Get ALL governance rules that apply to you: workflow constraints, assignment policies, behavioral requirements. Includes workspace and project-level rules with source annotations. Call at session start (ACP Step 3).

Get YOUR assigned tasks (ACP Step 5). Filter by status_category to focus on active work. Use at session start and after completing tasks to pick up the next assignment.

Get project details with statuses and custom fields.

Get ALL PERMANENT KNOWLEDGE for a project: decisions, conventions, accumulated context. Call at session start (ACP Step 2). Returns workspace-level + project-level memories. For RECENT events, use get_context instead.

Get all rules configured for a project (all scopes: workspace + project). Kept for backward compatibility — prefer get_my_rules for agent-scoped effective rules.

Returns the history of all instances for a recurring task schedule. ALWAYS call this when you receive a recurring task — it gives you context on what previous instances accomplished, what issues were found, and what artifacts were produced. Use it to continue work intelligently rather than starting from scratch.

Get full task details with optional comments, artifacts, dependencies, and VCS links.

Get EVERYTHING about ONE TASK in a single call: full details + comments + artifacts + dependencies + activity. Use when working on a specific task instead of calling get_task + list_comments + list_artifacts separately.

Get the workspace team directory listing all agents and human members with their profiles.

Get workflow rules for a project including allowed transitions, policies, and permissions for the calling agent.

Send heartbeat to stay visible. Call at session START with status=online, periodically during work with status=busy. Reports current_task_id, message, and metadata.

Import workspace configuration from YAML. Applies rules, statuses, and project templates defined in the YAML.

List artifacts attached to a task. Each carries download_path; a browser_only_url block, when present, is for a human and must not be fetched. Downloading an artifact is two GETs. Step 1: GET <base>/api/v1/artifacts/<id>/download with header X-Agent-Key: <your agent key> -> 200 JSON {"url": "<presigned URL>"}. Step 2: GET that url with NO headers -> 200, the file bytes. Pitfalls: on step 1 only X-Agent-Key is accepted (X-API-Key and Authorization: Bearer give 401); on step 2 any extra header, Authorization in particular, breaks the presigned signature (400). The artifact's download_path is step 1's path. Never fetch browser_only_url with an agent key: it is a human page and answers 401 by design.

List comments on a task. Paginated: call again with a higher `page` to read a thread longer than `limit`.

Read the comments on a document as threads — each top-level comment with its replies nested under it, the quoted passage it is anchored to, and who wrote it. Resolved threads are hidden unless include_resolved=true. A comment whose quoted text no longer exists in the document is marked orphaned=true in its anchor: it is still shown, and it is not pointing anywhere.

List a project's documents — id, title, slug path, version, who touched them last. Carries NO document bodies, so it is safe to call on a whole project: use it as the map, then get_doc for one page. Returns path and has_children for navigating the tree.

List available projects in the workspace.

Lists all recurring task schedules for a project.

List sub-agents of an agent.

List tasks with filters. Provide project_id for project-scoped listing or workspace_id for global search across all projects (requires search parameter). Each item's description is included by default and has_description always reflects the task's real content, computed before any trimming below — but on ANY page (plain listing or search=) whose descriptions total more than 200KB, the server blanks descriptions from the TAIL of that page (in item order) to keep the response size bounded, and marks the response truncated:true (field omitted when false). search= usually returns few enough hits to stay under that budget, so it is the practical workaround for a specific known task, but the one guaranteed way to read a given task's full description regardless of any listing's size or order is get_task(task_id).

Change task status (e.g. todo → in_progress → done). Use status SLUGS (not UUIDs). On move to 'review', task auto-reassigns to creator unless assignee_id is provided.

Long-poll for new task assignments. Blocks until a task is assigned to this agent or the timeout expires. Returns current assigned tasks and whether any change occurred. Kept for backward compatibility — prefer get_my_tasks for non-blocking access.

Publish an event to the event bus. For summaries, use event_type='summary'. Add memory={persist:true, key:'decision-name'} to also save as permanent memory. Replaces the deprecated publish_summary tool.

Publish a work summary event (convenience wrapper for publish_event with type=summary). Kept for backward compatibility — prefer publish_event with event_type='summary'.

SEARCH memory by keywords. Use to find a SPECIFIC piece of knowledge, e.g. 'API convention' or 'license decision'. Returns ranked results with scores. For loading ALL project knowledge at session start, use get_project_knowledge instead. Set include_archived=true to retrieve archived memories.

Search memory with Knowledge Graph expansion. Seeds from hybrid recall, then BFS-traverses memory_edges up to hops depth. Returns memories ranked by composite score with hop_distance and provenance fields. Use when you want broader context — related decisions, connected incidents, derived learnings.

Record a directive from the workspace owner as a canonical decision in project_knowledge. Broadcasts to specified agents via propagate_to tags. privacy:private records are stored but EXCLUDED from get_canonical_updates. Auto-flags private if text contains secrets. If task_id is given, also records this as a human_gate decision on that task (docs/human-gate-decision-recorded.md in evc-mesh) — releases the gate as a consequence if it's currently live, and links back via canonical_key. Best-effort: a failure here is reported in the result but does not undo the canonical write.

Register a sub-agent under the calling agent.

Release the exclusive lock on a task acquired via checkout_task. Call when done with the task or if you need to hand it off. The lock is also released automatically when it expires.