fix: close Action exfiltration channel + wire CVE toggle (audit H1/M1/L2) - #50
Merged
Conversation
… (audit H1/M1/L2) - H1/L2: the narrate job re-opened the exact exfiltration channel the two-job split existed to close — it fetched only the skill .md (not the CLI), so the CLI-first skill hit its LLM fallback and RE-READ the raw untrusted repo, while holding ANTHROPIC_API_KEY; and 'no egress' was false because the reduced toolset still had Bash (curl). Fix: render the report DETERMINISTICALLY via cli/dsgai_report.py — no LLM, no API key (secret removed from the workflow entirely), no repo re-read. Also fixes L2 (the deterministic renderer was unused). The job is now secretless and safe on forks. Verified: render-only with just report.py + css + checkpoint produces the HTML, no secret leak. - M1: run_cve_enrichment input is now wired to --no-cve (was dead; CVE always ran). actionlint + yamllint clean; ANTHROPIC_API_KEY no longer referenced anywhere.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
H1 (critical Action finding): the
narratejob re-opened the exfiltration channel the two-job split exists to close — it fetched only the skill.md(not the CLI), so the CLI-first skill hit its LLM fallback and re-read the raw untrusted repo while holdingANTHROPIC_API_KEY; and 'no egress' was false because the reduced toolset still had Bash (curl).Fix: render the report deterministically via
cli/dsgai_report.py— no LLM, no API key (the secret is now absent from the entire workflow), no repo re-read. The job is secretless and safe on forks. This also fixes L2 (the deterministic renderer was unused). Verified: render-only with justreport.py+css+checkpoint produces the HTML, no secret leak.M1:
run_cve_enrichmentis now wired to--no-cve(was a dead input; CVE always ran).actionlint+yamllintclean;ANTHROPIC_API_KEYno longer referenced.