| name | dcr-reviewer-init |
| description | Join Deep Code Review as a reviewer for an active review run. |
| invocation-args | {"task_name":"required","suffix":"required","domain":"optional"} |
dcr-reviewer-init
Do not start this reviewer role implicitly.
Bind the reviewer run to an explicit task_name and suffix at invocation time.
Add domain when the human or lead assigns one.
Read First
.agent/skills/code-review/SKILL.md
.agent/playbooks/deep-code-review/PLAYBOOK.md
.agent/playbooks/deep-code-review/template/round_template/README.md
Keep these hot-loaded for the duration of the playbook:
.agent/skills/code-review/SKILL.md
.agent/playbooks/deep-code-review/skills/dcr-reviewer-init/SKILL.md
For every new round, begin by reading the active round README.md.
Then follow 00_round_context.md for the full round contract and read order.
If this is not the first round, also read the prior round outputs referenced in 10_previous_round_feedback.md.
Identity Contract
Before writing findings, know:
- the active task name
- your role in the roster
- your reviewer suffix / filename
- the exact round directory you are reviewing from
Your output file should be:
reviewers/reviewer_<suffix>.md
Do not edit findings_ledger.md.
Finding ids in shared artifacts are maintained by the lead only.
Your Job
You are responsible for:
- reviewing the change using
.agent/skills/code-review/SKILL.md
- verifying the live file state before writing findings
- validating the status of prior findings you raised
- reusing existing finding ids when the round artifacts provide them
- avoiding repetition of rejected findings without new evidence
- writing a structured findings file that the lead can ingest
Live Review Rule
The active round packet is the live source of truth.
Do not infer scope, reviewed state, or reviewer target paths from chat if the round artifacts already define them.
Do not duplicate or override the active round packet with ad hoc reviewer instructions unless the human explicitly changes the round plan.
If You Have No Findings
Do not drop from the review automatically.
If you reach No findings.:
- write that explicitly
- stay in the roster
- move into lightweight follow-up mode
Lightweight follow-up mode means:
- validate prior concerns were fixed or deferred correctly
- watch for regressions introduced by new fix batches
- avoid re-running a heavy cold review unless asked