aws-samples / aws-samples/sample-autonomous-cloud-coding-agents
feat(ci): pin Semgrep + rule packs (semgrep = "latest" + --config auto lets registry changes break CI)
- Dominant language
- TypeScript
- Stars
- 143
- Forks
- 46
- Avg merge
- 3d 9h
- Merged PRs (30d)
- 20
Description
## Problem
`mise.toml` pins `semgrep = "latest"` (line 27) and `security:sast` uses `--config auto` (line 140), which fetches rules **live from the semgrep registry** at scan time. This means CI's SAST behavior can change with no code change on our side — a newly published or modified rule can turn `main` red unexpectedly.
This is not hypothetical: it is the **root cause of #532** — the `uv-missing-dependency-cooldown` rule (added in semgrep 0.9.17) was picked up automatically and failed the suite, with no change to our code.
## Proposed
- Pin the semgrep engine to an exact version (not `latest`) in `mise.toml`.
- Pin / vendor the rule packs (or at least the `--config` sources) so rule changes are adopted deliberately, not silently.
- Consider a periodic bump (like `upgrade-main.yml`) so updates are reviewed in a PR rather than surprising a scheduled run.
## Acceptance criteria
- SAST rule set is version-controlled / pinned; a registry-side rule change cannot change CI outcome without a PR.
## Context
Maintainer-requested follow-up from #532. See also related CI-hardening issues #329, #328.
Contributor guide
Research direction
Start with mise.toml lines 27 and 140, then inspect the CI workflow and upgrade-main.yml referenced in the issue. Review #532 and the related CI-hardening issues to understand the failure and existing update patterns. Done means registry-side Semgrep rule changes cannot alter SAST results without a reviewed PR.
Written by the indexing model from the issue text.
Assessment
- Domain
- ci-cd, devops, security
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 52/100