Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add ccheney/robust-skills --skill bazelgit clone --depth 1 https://github.com/ccheney/robust-skillsWrote 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/ccheney/robust-skills/bazel)<a href="https://agentmods.dev/skills/ccheney/robust-skills/bazel"><img src="https://agentmods.dev/badge/skills/ccheney/robust-skills/bazel/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/ccheney/robust-skills/bazel"><img src="https://agentmods.dev/badge/skills/ccheney/robust-skills/bazel.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00041 | $0.00769 |
| Opus 5.5 | $0.00016 | $0.00308 |
| Sonnet 5.5 | $0.00008 | $0.00154 |
| Haiku 4.5 | $0.00004 | $0.00077 |
Grade A, and why
bazel 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 1mo 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.
How it starts
The opening of the file, as written. The whole thing — 53 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Bazel Monorepos
Make build changes against the actual target graph and pinned Bazel/ruleset versions. Examples cover Bazel 9.2.0; they are not an upgrade instruction. The user's explicit instructions take precedence over this skill's guidelines.
Establish the relevant context
Use the affected targets, error output, or requested build outcome as input. Check .bazelversion, relevant module declarations, BUILD files, and the configuration involved in the task. Read other repository configuration only when it affects that question.
| Question | Tool |
|---|---|
| Target dependencies, reverse dependencies, paths | bazel query |
Selected flags, platforms, toolchains, or select() branches |
bazel cquery with the failing invocation's configuration |
| Actual commands, inputs, outputs, tools | bazel aquery |
| External modules, versions, generated repositories | bazel mod |
| Cache misses or execution performance | Execution logs, profiles, BEP |
If the failing target or invocation cannot be inferred, ask for it and continue independent inspection. Do not invent repository pins or remote-cache endpoints.
Preserve the build contract
- Declare direct inputs, dependencies, tools, and runtime data on the owning target. Use runfiles APIs for runtime resources.
- Keep package boundaries and visibility deliberate. Neither one target per file nor one per service is a universal template.
- For Bazel 9, use Bzlmod and explicit external language-rule loads. For older repositories, follow version-matched rules until an upgrade is requested.
- Diagnose missing inputs or configuration differences before using
clean --expungeor disabling sandboxing; those can hide the cause. - Validate affected targets and configurations. Reserve graph-wide builds/tests for graph-wide impact or required repository checks.
Deliver the requested patch or diagnosis with relevant target labels, evidence, and unresolved configuration dependencies. Finish checks needed for that outcome without expanding a local fix into a repository migration.
What ships with it
10 files 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.
- agents/openai.yaml 192 B
- references/ADOPTION.md 16 KB
- references/BUILD-GRAPH.md 15 KB
- references/BZLMOD.md 19 KB
- references/CHEATSHEET.md 15 KB
- references/CI-PERFORMANCE.md 19 KB
- references/CONFIGURATION.md 16 KB
- references/HERMETICITY-CACHING.md 19 KB
- references/QUERIES-DEBUGGING.md 18 KB
- references/STARLARK.md 18 KB
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.
- 1mo ago Changed · -326 lines · -89 tokens per session 806f51e78f1e
- 1mo ago First seen · 379 lines · 130 tokens per session scan A 41ffc29b6427
bazel is a skill published in the GitHub repository ccheney/robust-skills (61 stars, last pushed 21d ago), licensed MIT. It adds 41 tokens to every session and 769 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
workflow-branch-sync
Sync the current branch with the default branch via rimba or git, then resolve conflicts file by file with a per-hunk rationale and a keep-mine / keep-main / manual recommendation before staging. Push only happens after explicit confirmation — never automatic.
workflow-cleanup-merged
Use after a PR has been merged on GitHub to remove the local worktree, delete the local branch, delete the remote branch, and fast-forward local main — safely, with squash-merge support.
workflow-pr-review
Use when reviewing a remote GitHub PR — a first-pass peer review, or a followup re-check after the owner has addressed feedback. Fetches into an ephemeral worktree, runs the reviewer agent against the updated diff with a Review Decision footer instruction, deduplicates findings against existing review threads (±5-line…
workflow-address-feedback
Use when a PR owner wants to address review feedback — fetches outstanding threads and general comments, presents a per-item triage (ADDRESSED / CLARIFIED / DEFERRED), applies fixes via the Edit tool, commits via workflow-commit-and-pr, resolves review threads via GraphQL resolveReviewThread (those without a thread…
workflow-commit-and-pr
Use when the user wants to commit staged changes or create a PR — enforces trigger-phrase discipline (preview vs commit vs ship), the [type] commit format, branch-naming check, draft-vs-ready PR prompt, and PR template detection. Pre-merge counterpart to workflow-cleanup-merged.
workflow-development
Development workflow — full lifecycle from Branch → Implement → Verify → Review → Deliver. Activated by /swe-workbench:implement, /swe-workbench:design, /swe-workbench:refactor, /swe-workbench:debug, /swe-workbench:test, /swe-workbench:architect, /swe-workbench:migrate, and /swe-workbench:document when the plan being…