Problem
The static-analysis workflow is filing the same RGS-* security findings each day, recreating issues that were closed the previous day:
| Rule |
2026-05-07 (closed) |
2026-05-08 (refiled) |
| RGS-004 (comment-triggered workflow w/o auth) |
#30778 |
#30945 (brave.lock.yml) |
| RGS-012 (secret exfil via outbound HTTP) |
#30776 |
#30947 (visual-regression-checker.lock.yml) |
| RGS-018 (suspicious payload exec pattern) |
#30777 |
#30946 (api-consumption-report.lock.yml) |
The rule IDs are identical; only the affected files differ. The workflow either (a) treats every file independently with no awareness of prior closure, or (b) lacks a fingerprint that matches across runs.
Suggested Fix
In the static-analysis workflow prompt:
- Before filing a
create_issue, search existing issues by title pattern [static-analysis] RGS-NNN and the file path; if a CLOSED issue exists for the same rule+file, skip; if an OPEN issue exists, add a comment instead.
- Optionally include a
rule + file-hash fingerprint in a hidden HTML comment to make matching robust.
Suggested Agent
agentic-workflows — workflow .md prompt edit.
Estimated Effort
1–2 hours.
Source
DeepReport 2026-05-08.
Generated by DeepReport - Intelligence Gathering Agent · ● 18.1M · ◷
Problem
The
static-analysisworkflow is filing the same RGS-* security findings each day, recreating issues that were closed the previous day:brave.lock.yml)visual-regression-checker.lock.yml)api-consumption-report.lock.yml)The rule IDs are identical; only the affected files differ. The workflow either (a) treats every file independently with no awareness of prior closure, or (b) lacks a fingerprint that matches across runs.
Suggested Fix
In the static-analysis workflow prompt:
create_issue, search existing issues by title pattern[static-analysis] RGS-NNNand the file path; if a CLOSED issue exists for the same rule+file, skip; if an OPEN issue exists, add a comment instead.rule + file-hashfingerprint in a hidden HTML comment to make matching robust.Suggested Agent
agentic-workflows— workflow .md prompt edit.Estimated Effort
1–2 hours.
Source
DeepReport 2026-05-08.