Default policy: record everything, ask a person only for a short list of costly actions; today's rules as strict mode
Maintainers usually reply within 1 day
Nobody has claimed this yet.
Assessment
- Difficulty
- 5/5
- Estimated time
- Over a week
- Newbie friendliness
- 25/100
- Issue type
- Feature
- Clarity
- Mostly clear
- Activity status
- Active
- Tech stack
- python
- Domain
- ai-infra-agents, security
Research direction
Start by reading the policy rules and replay fixtures behind #598, then identify the E17 and E21 baselines and the viewer, doctor, README, and docs/limits.md entry points. Replay the owner's week against the proposed default and strict mode; done means all merge-gate counts and advisories pass, strict verdicts match today's baseline, and the docs and viewer report the new behavior.
Written by the indexing model from the issue text.
Description
On a coding agent's real week, Tracekit still stops about 29.6 harmless calls a day, down from 43 after #598. Every harmless stop teaches the person to approve without reading. The owner approved a new default on 2026-11-04.
The new default (coding agents): every call is still decided, signed and recorded. A person is asked only for a short list of costly or security-relevant actions:
- spending money;
- sending messages, emails or posts outside;
- deleting data outside the project or on remotes;
- deploy, push or merge to shared places;
- touching secrets or credentials, or sending them out;
- running downloaded code;
- changing the guardrails (hooks, Tracekit, agent harness settings).
Lines Tracekit can't read still wait for the person. Everything else is recorded and allowed.
Strict mode: today's full rules, as one switch (tracekit init --strict, or a pack name). Strict is recommended for production servers and evals.
Done when:
- A replay of the owner's real week through the default meets the merge gate:
- all 8 needed stops still stop;
- under 5 harmless stops a day;
- every spelling from the published advisories, and from #598's review rounds, still stops where it touches the short list.
- Strict mode gives today's verdicts on E17 and E21, with the baseline unchanged.
- The README,
docs/limits.mdand the demo describe the default honestly, andtracekit doctorsays which mode is active. - The viewer shows a "worth a look" list per run: the recorded calls that matched a former stop rule.
- Dominant language
- Python
- Stars
- 137
- Forks
- 8
- Avg merge
- 4h 39m
- Merged PRs (30d)
- 376
Getting set up
- No Dockerfile or Docker Compose file
- Has a pull request template
- Read the contributing guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from Cygnux-Labs/Tracekit
-
MCP client: record `get_prompt` on a wrapped session as `mcp:<server>/prompt:<name>`Possibly taken @Mr45401 claimed this 1 day ago. Opengood first issue help wanted
Difficulty 2/5 1-3 hours Newbie friendliness 85/100
Cygnux-Labs/Tracekit#666 · 1 comment ·
Maintainers usually reply within 1 day
-
good first issue help wanted
Difficulty 2/5 1-3 hours Newbie friendliness 84/100
Cygnux-Labs/Tracekit#614 ·
Maintainers usually reply within 1 day
-
good first issue help wanted recurring
Difficulty 2/5 1-3 hours Newbie friendliness 66/100
Cygnux-Labs/Tracekit#387 · 1 comment ·
Maintainers usually reply within 1 day
-
good first issue
Difficulty 2/5 1-3 hours Newbie friendliness 76/100
Cygnux-Labs/Tracekit#259 · 5 comments ·
Maintainers usually reply within 1 day
-
good first issue help wanted
Difficulty 4/5 3-5 days Newbie friendliness 55/100
Cygnux-Labs/Tracekit#672 ·
Maintainers usually reply within 1 day
All issues in Cygnux-Labs/Tracekit
Similar issues
-
Difficulty 2/5 1-3 hours Newbie friendliness 67/100
bigbio/sdrf-pipelines#372 · 1 comment ·
Maintainers usually reply within 1 day
-
Difficulty 1/5 Under an hour Newbie friendliness 84/100
521xueweihan/HelloGitHub#4297 ·
-
Difficulty 2/5 1-3 hours Newbie friendliness 83/100
aws-samples/sample-per-model-bedrock#822 ·
Maintainers usually reply within 1 day
-
area:study-data p2-high ready-for-agent type:chore
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
fderuiter/deruiter.dev#3706 ·
Maintainers usually reply within 1 day
-
bug
Difficulty 2/5 1-3 hours Newbie friendliness 74/100
awslabs/visual-asset-management-system#1234 ·
Maintainers usually reply within 1 day