# Research Thread: [Run ID]

## Mission

- Objective:
- Scope:
- Started:
- Owner:
- Current status: active / paused / complete / abandoned

## Locked Surfaces

List files or rules the agent may read but not mutate during this run.

- `researcher/rubrics/content-curation.md`
- `researcher/rubrics/skill-change.md`
- `researcher/rubrics/harness-change.md`
- `researcher/mechanisms/registry.jsonl`
- `.claude-plugin/marketplace.json`
- `.plugin/plugin.json`
- Merge policy: agents may prepare PRs, but human approval is required for push and merge.

## Editable Surfaces

List files the agent may create or update during this run.

- Source evaluations:
- Skill proposals:
- Mechanism proposals:
- Run state:
- Draft branches or patches:

## Source Queue

| ID | Source | Status | Next Action |
| --- | --- | --- | --- |
| S001 |  | discovered | evaluate |

## Decisions

Append decisions chronologically.

```text
T+00:00 decision: [what changed]
reason: [why]
evidence: [file or URL]
next: [next action]
```

## Experiments And Evaluations

| ID | Artifact | Rubric | Result | Notes |
| --- | --- | --- | --- | --- |
| E001 |  |  |  |  |

## Open Questions

- 

## Handover Summary

Use this section whenever context compacts or another agent resumes the run.

- Best current candidate:
- Rejected candidates:
- Unresolved risks:
- Files to read first:
- Next concrete action: