Tracely is a CI/CD system for AI agents that turns failed production traces into replayable regression tests. Development teams use it to detect and group agent failures, run the resulting cases on pull requests, and block changes that reproduce those failures. The catalogue entries provide skills for operating this trace-based testing and observability workflow.
Borrowing it
Nothing to install: this file belongs to Jwuthri/Tracely-ai. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/Jwuthri/Tracely-ai/master/.claude/skills/papercuts/SKILL.mdgit clone --depth 1 https://github.com/Jwuthri/Tracely-aiWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/jwuthri/tracely-ai/papercuts)<a href="https://agentmods.dev/skills/jwuthri/tracely-ai/papercuts"><img src="https://agentmods.dev/badge/skills/jwuthri/tracely-ai/papercuts/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/jwuthri/tracely-ai/papercuts"><img src="https://agentmods.dev/badge/skills/jwuthri/tracely-ai/papercuts.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00115 | $0.00949 |
| Opus 5 | $0.00057 | $0.00475 |
| Sonnet 5 | $0.00023 | $0.00190 |
| Haiku 4.5 | $0.00012 | $0.00095 |
Grade A, and why
papercuts scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
This is a copy
100% identical to papercuts — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 82 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Papercuts
Capture small friction in the moment without derailing the current task. Aggregated entries show where the repository needs sanding down — so the bar is that a different contributor would hit the same thing, and the repository can do something about it.
The two-question test
Log it only if both are true:
- Reproducible for anyone. A different person, on a fresh checkout, working in this repo would hit the same friction. It is not specific to your sandbox, shell config, machine, network, or a one-time hiccup.
- Fixable in the repo. A change to the repo's code, config, scripts, or docs would prevent or reduce it.
If either answer is "no," push through it and move on — do not log it.
Do NOT log
- Your environment's failures. Sandbox
EPERM/listen/ IPC-socket errors, blocked network orfetch failed, permission denials, missing system tools. That is the runner, not the repo. - Your own shell mistakes. Reserved or special variable names (
status,path), unquoted globs, a broken login-shell hook. Fix the command — there is nothing in the repo to sand down. - Transient flakiness. A command that succeeded on retry with no repo-side cause (a network blip, a hung push, a slow mirror).
- Local state you corrupted. A partial
node_modulesafter branch-switching, a stale dev-server port, a dirty cache. Re-run the install or cleanup. - Third-party or beta-tool limitations the repo can't change — unless the fix is a repo-side workaround worth writing down (then log that workaround).
- Product or code correctness bugs (fix now or track as real work), and what you accomplished (that belongs in the task summary).
- Secrets, credentials, personal data, raw customer payloads, or sensitive paths.
When something fails, first ask "is this the repo, or is this me/my environment?" Only the former is a papercut.
Log proactively
- Search
.agents/PAPERCUTS.mdfor an equivalent entry and avoid duplicates. - Append one unchecked item under
## Openusing this format:
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 9d ago First seen · 82 lines · 115 tokens per session scan A 9cabafbd45f1
papercuts is a skill published in the GitHub repository Jwuthri/Tracely-ai (1,216 stars, last pushed yesterday), licensed MIT. It adds 115 tokens to every session and 949 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. It is 100% identical to papercuts, differing in 0 lines, and is treated as a copy.
Other skills, from other repositories
architect-analyze
Analyze architecture for consistency between ADRs and AD, completeness, and quality issues. Use when validating generated or refined architecture artifacts, before feature development, during architecture review, or periodically to detect drift.
squid-implement-night
Run the full agent-team pipeline end-to-end for one feature whose Tasks Plan is already approved by /squid-plan, handing the human a validated, ready-to-squash-merge PR. Trigger after /squid-plan.
tech-debt-ci-review
Codex adapter for deep technical-debt and CI-stability audits. Use when asked to find test theater, flaky tests, missing or mis-scoped tests, brittle CI/toolchain behavior, structural debt blocking green PRs, or a remediation order for opencode-swarm.
gemini-review
Google Gemini CLI code review with Gemini 2.5 Pro, 1M token context, CI/CD integration.
aster-review-ci
Run aster code reviews non-interactively in CI, GitHub Actions, or from another agent. Covers aster review --pr, --json, --stream, --comment, diff-from-stdin, token handling, and filtering findings. Use when wiring aster into a pipeline, posting PR comments, or parsing review output programmatically.
review-deep
Review change vs spec, plan, code quality heuristics, and repo rules; append remediation phase if needed.