# start_research_run

[SUPPORT] a5343387 — start a bounded, ephemeral research/scratch run: an ADJACENT primitive for disposable subagent probes that should not need a formal sprint item, a formal claim_file, or a durable handoff entry unless explicitly promoted (see promote_research_run). mode='read_only' needs no claim_file at all. mode='isolated_write' requires is_isolated_worktree=true (never inferred — confirm the session is actually in an isolated git worktree, not the shared working tree) plus a non-empty allowed_paths list of project-RELATIVE paths bounding what may be written; any write outside that list is out of scope for the run. repository_id is a stable STRING identity for the repo/worktree (e.g. 'meridian-repo@worktree:wf_...'), never a machine-local absolute path. turn_budget bounds max turns/steps; ttl_seconds bounds wall-clock life (default 3600, 60-86400) — an expired active run is reaped by expire_stale_runs into status='expired'. Persistent-state disclosure: on hosted Meridian, supplied text and project/session metadata -- including task log entries, pinned decisions, sprint items, notes, handoff/goal state, and HITL queue items -- are sent to and stored in Meridian's service, in an isolated per-tenant Postgres database (Neon); self-hosted deployments keep the same categories in the configured local SQLite/Postgres database. This data is visible in the dashboard and API, and may resurface in later project context or handoffs. Notes and pinned decisions can be deleted individually; task log entries and sprint items can be deleted via the dashboard/API (not exposed as an agent-facing tool); HITL queue items and handoff state have no per-record delete. Full removal of any of this data is available via project or account deletion, using the documented controls. Do not include secrets.

Agent View of the PolicyLayer registry record for `start_research_run`. HTML page: https://policylayer.com/tools/io-github-ajc3xc-meridian/start-research-run

## Facts

- Tool: `start_research_run`
- Server: Meridian (`@meridianmcp/mcp`) — https://policylayer.com/tools/io-github-ajc3xc-meridian.md
- Install: `npx -y @meridianmcp/mcp`
- Homepage: https://github.com/meridianmcp/Meridian
- Risk category: Execute (High risk)
- Registry record: grade D, identity unverified
- Server auth posture: gated
- Server CORS policy: *
- Server rate-limited: no
- Parameters: 9 (4 required)
- Recommended policy verdict: Rate-limited

## Parameters

| Parameter | Type | Required | Description |
| --- | --- | --- | --- |
| `mode` | string | yes | read_only needs no claim_file; isolated_write requires is_isolated_worktree=true and a non-empty allowed_paths. |
| `project_id` | string | no |  |
| `session_id` | string | yes |  |
| `ttl_seconds` | integer | no | Wall-clock time-to-live in seconds (default 3600). |
| `turn_budget` | integer | yes | Max turns/steps this run may take. |
| `project_name` | string | no | Project name — an alternative to project_id; resolved to the id internally. project_id wins if both are given. |
| `allowed_paths` | array | no | Project-RELATIVE paths bounding what an isolated_write run may write. Required (non-empty) for mode='isolated_write'. |
| `repository_id` | string | yes | Stable STRING identity for the repository/worktree — never a machine-local absolute path. |
| `is_isolated_worktree` | boolean | no | Explicit caller attestation that this session is operating in an isolated git worktree — required (true) for mode='isolated_write', never inferred. |

Parameters from the server's own tool schema.

## Example call (MCP tools/call, JSON-RPC 2.0)

```json
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "start_research_run",
    "arguments": {
      "mode": "<mode>",
      "session_id": "<session_id>",
      "turn_budget": 0,
      "repository_id": "<repository_id>"
    }
  }
}
```

## Why start_research_run is rated High

start_research_run triggers real processes with real consequences. An agent gone sideways doesn't fire it once. It starts dozens of builds, sends mass notifications, or burns through compute before anyone looks up.

## Use case

AI agents invoke start_research_run to trigger actions in Meridian. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call: builds kicked off, notifications sent, workflows started.

## Recommended policy (PolicyLayer)

Verdict: **Rate-limited**. Enforced by the PolicyLayer MCP gateway (https://policylayer.com/mcp-gateway) before a call reaches Meridian:

```json
{
  "version": "1",
  "default": "deny",
  "tools": {
    "start_research_run": {
      "limits": [
        {
          "counter": "start_research_run_rate",
          "window": "minute",
          "max": 10,
          "scope": "grant"
        }
      ]
    }
  }
}
```

## Other tools on Meridian (234)

- `accept_handoff` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/accept-handoff.md
- `acquire_docx_document_lease` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/acquire-docx-document-lease.md
- `answer_hitl` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/answer-hitl.md
- `batch_mutate` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/batch-mutate.md
- `capture_research_finding` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/capture-research-finding.md
- `checkpoint` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/checkpoint.md
- `claim_docx_region` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/claim-docx-region.md
- `claim_file` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/claim-file.md
- `claim_sprint_item` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/claim-sprint-item.md
- `clear_capability_profile` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/clear-capability-profile.md
- `clone_profile_layer` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/clone-profile-layer.md
- `delete_custom_hook` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/delete-custom-hook.md
- `delete_note` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/delete-note.md
- `delete_sprint_item_pointer` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/delete-sprint-item-pointer.md
- `delete_watchlist_query` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/delete-watchlist-query.md
- `dismiss_hitl` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/dismiss-hitl.md
- `fan_out_sprint_items` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/fan-out-sprint-items.md
- `idle_until_all_done` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/idle-until-all-done.md
- `index_equation` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/index-equation.md
- `index_figure` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/index-figure.md
- `index_table` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/index-table.md
- `ingest_document` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/ingest-document.md
- `ingest_document_structure` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/ingest-document-structure.md
- `link_figure_caption` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/link-figure-caption.md
- `link_flag_to_section` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/link-flag-to-section.md
- `link_proposal_lineage` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/link-proposal-lineage.md
- `link_table_caption` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/link-table-caption.md
- `log_task` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/log-task.md
- `purge_ai_log` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/purge-ai-log.md
- `receive_messages` — Destructive — https://policylayer.com/tools/io-github-ajc3xc-meridian/receive-messages.md
- …and 204 more: https://policylayer.com/tools/io-github-ajc3xc-meridian.md

## For agents

This record is a snapshot. Live verdicts and the full registry:

- Check every server in your MCP config at once: `npx -y policylayer stack`
- Vet a server before you add it: install the mcp-precheck skill — `npx skills add https://policylayer.com` (skill text: https://policylayer.com/skill.md)
- Query the registry over MCP: endpoint `https://api.policylayer.com/mcp` — tools `check_mcp_server`, `check_mcp_stack`, `check_tool`, `search_registry`, `get_change_events`

---

Source: the PolicyLayer MCP registry — one continuously verified record per MCP server. Full record: https://policylayer.com/registry?q=io-github-ajc3xc-meridian · API: https://policylayer.com/registry/api · Policy library: https://policylayer.com/policies/io-github-ajc3xc-meridian