Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Get Started Free →

PR / Issue Triage · mkurman | DecimalAI

mkurman/pr-issue-triage

SkillScore100Maintained

Canonical PR/Issue triage pack combining repo review queues with work tracker context and approval-gated write-backs.

productivityby mkurman

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Project Management pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/mkurman-pr-issue-triage/SKILL.md

decimalai skills pull mkurman-pr-issue-triage --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/mkurman-pr-issue-triage/SKILL.md https://app.decimal.ai/s/mkurman-pr-issue-triage/SKILL.md

Raw file: /s/mkurman-pr-issue-triage/SKILL.md ↗ · machine-readable summary at /s/mkurman-pr-issue-triage.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 8d ago

Yes — +32% lift+1% tokens0% turns100% of whole cases pass with skill (22/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-04	✗→✓	▲ Improved	3%	0%
case-06	✗→✓	▲ Improved	-36%	0%
case-07	✗→✓	▲ Improved	-21%	0%
case-12	✗→✓	▲ Improved	-35%	0%
case-13	✗→✓	▲ Improved	-13%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

mkurman / zorai

skills/workflow-packs/pr-issue-triage/SKILL.md

View on GitHub ↗

License MITImported 8/3/2026Version v1

Skill contents

SKILL.md

PR / Issue Triage

User story

I want one triage view across repo review items and issue tracker work, so I can see stale, blocked, and actionable items without manually stitching together GitHub/GitLab and Linear/Jira.

Pack contract

Prerequisites and readiness

Requires one repo connector: github or gitlab
Optionally enriches with one tracker connector: linear or jira
If the tracker is unavailable, the pack still runs as repo-only triage and clearly marks tracker enrichment as degraded.

Inputs and configuration fields

repo_connector: github or gitlab
tracker_connector: linear, jira, or none
owner / repo or equivalent project coordinates
include_writeback_suggestions: boolean
stale_days: integer threshold

Outputs and delivery targets

merged queue showing:
open PRs / review items
open issues / work items
stale items
blocked items
suggested next actions
source links should point back to the original PR/issue/tracker item
write-back suggestions must be clearly marked approval-required

Manual run recipe

List repo review items and work items from the chosen repo connector.
If tracker is ready, fetch tracker work items and merge by title/ID/reference where possible.
Summarize stale and blocked items.
Suggest write-backs separately from the read-only report.

Example routine wiring

Schedule as a weekday triage routine or on-demand review block. Example payload description:

Run the canonical PR/Issue Triage pack for the main repo, merge repo/tracker context, and surface approval-gated write-back suggestions.

Example prompt

Run PR/Issue Triage for owner/repo using GitHub plus Linear. Show stale PRs, blocked issues, and suggested labels/comments, but do not perform write-backs without approval.

Failure and recovery behavior

Missing repo connector -> fail closed with setup hint.
Missing tracker connector -> continue with repo-only triage.
Proposed write-back with no approval -> keep suggestion in the report, do not execute it.

Verification checklist

] Manual repo-only proof passes.
] Manual repo + tracker proof passes.
] Missing tracker degrades gracefully.
] Write-back suggestions are clearly approval-gated.
] Source links are present for triaged items.

Want to see if mkurman/pr-issue-triage would activate on your phrasing? Try it before you install — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/21/2026, 17:53 UTC · 22 cases

Lift

+32%

100% vs 68%

Agent turns

0%

1.0 vs 1.0

Tokens

+1%

1,874 vs 1,851.91

fail → pass · 7pass → pass · 15

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-03 Triage the repo acme/frontend using GitHub and Linear. Several stale PRs need closing tags and labeling. The user has not granted pre-approval or explicit permission for modifications. Perform the triage.	pass→pass	10,255	11,703	+14%	1	1	0%	1,584	2,607	+65%	0	0	—
case-04 Create a unified triage view for repo enterprise/backend combining GitHub PRs and Jira issues. How should individual work items be referenced in the merged queue?	fail→pass	18,886	11,595	-39%	1	1	0%	2,456	2,536	+3%	0	0	—
case-14 What key status conditions must be extracted and summarized when scanning repository review items and tracker tickets?	pass→pass	12,563	7,134	-43%	1	1	0%	1,945	1,651	-15%	0	0	—
case-01 Run a triage check on the octocat/Hello-World project. No GitHub or GitLab repository credentials or connectors are configured in the environment, but a Jira connector is present. Provide the triage report.	pass→pass	12,836	63,712	+396%	1	1	0%	1,473	1,242	-16%	0	0	—
case-02 Run a triage check for owner/repo using the GitHub connector. A Linear connector was requested but no API token or connector is configured for Linear. Generate the report.	pass→pass	11,280	10,904	-3%	1	1	0%	1,178	2,573	+118%	0	0	—
case-05 Assess open pull requests and work items for workspace-org/mobile-app using GitHub. We consider items inactive for 14 days to be stale. Detail how inactivity filtering should be parameterized.	pass→pass	19,095	15,793	-17%	1	1	0%	2,887	3,304	+14%	0	0	—
case-06 Attempt to run PR/issue triage on a Bitbucket repository `org/repo-name` with Jira connected.	fail→pass	13,598	6,234	-54%	1	1	0%	2,219	1,426	-36%	0	0	—
case-07 Configure a triage run for a team that only uses GitLab for both merge requests and issue tracking, without any external issue tracker like Linear or Jira.	fail→pass	19,060	10,203	-46%	1	1	0%	2,950	2,316	-21%	0	0	—
case-08 Generate a comprehensive triage report structure for team-alpha/service using GitHub and Linear. List the essential item categories that must be presented in the consolidated view.	pass→pass	18,694	16,671	-11%	1	1	0%	2,884	3,362	+17%	0	0	—
case-09 When compiling a unified triage summary for octocat/Spoon-Knife using GitHub, what recommendations should accompany detected stale or blocked pull requests?	pass→pass	14,050	9,478	-33%	1	1	0%	2,257	2,063	-9%	0	0	—
case-10 Configure a triage pipeline for group/project on GitLab combined with Jira Cloud.	pass→pass	17,076	14,205	-17%	1	1	0%	2,821	2,898	+3%	0	0	—
case-11 Configure a triage pipeline for company/core-api on GitHub combined with Linear.	pass→pass	17,156	10,866	-37%	1	1	0%	2,827	2,342	-17%	0	0	—
case-12 What schema field controls whether write-back recommendations are included when preparing a triage run for owner/repo?	fail→pass	8,240	1,817	-78%	1	1	0%	1,256	821	-35%	0	0	—
case-13 During an automated run of triage on repo/service, 3 stale issues were identified for closure. Should the pipeline automatically post the closure API call?	fail→pass	9,081	4,114	-55%	1	1	0%	1,361	1,182	-13%	0	0	—
case-15 How should suggested write-back actions (like adding labels or closing PRs) be presented relative to the read-only triage audit report?	pass→pass	17,407	8,914	-49%	1	1	0%	2,510	2,044	-19%	0	0	—
case-16 When both GitHub pull requests and Linear tickets are retrieved during triage, how are corresponding items linked or merged into single entries?	pass→pass	12,928	6,521	-50%	1	1	0%	1,985	1,517	-24%	0	0	—
case-17 List the valid allowed options for the repository connector setting in this triage tool.	fail→pass	2,353	1,721	-27%	1	1	0%	281	740	+163%	0	0	—
case-18 List the valid allowed options for the issue tracker connector setting in this triage tool.	fail→pass	2,353	2,192	-7%	1	1	0%	310	798	+157%	0	0	—
case-19 What project identification parameters are required to locate repository artifacts when initiating triage?	pass→pass	12,031	2,313	-81%	1	1	0%	2,064	926	-55%	0	0	—
case-20 Write a step-by-step GitHub Actions workflow YAML file that triggers on `pull_request` events and runs `actions/checkout@v4` followed by a custom bash script to auto-merge PRs.	pass→pass	11,087	9,871	-11%	1	1	0%	2,083	2,355	+13%	0	0	—
case-21 Write a Jira Query Language (JQL) query to find all unresolved issues in project 'PROJ' updated more than 30 days ago assigned to 'johndoe'.	pass→pass	3,654	3,483	-5%	1	1	0%	639	1,146	+79%	0	0	—
case-22 Write a GraphQL mutation for the Linear API to update an issue's state ID to 'done-state-id' given an issue ID 'issue-123'.	pass→pass	4,793	4,490	-6%	1	1	0%	772	1,379	+79%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +32 percentage points is the difference between those two pass rates over the 22 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 68.2%
Pass rate with the skill: 100%
Cases attempted: 22
Cases counted toward lift: 22
Cases that regressed: 0
Skill version measured: v1
Run completed: 8/21/2026, 17:53 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-04

Prompt

Create a unified triage view for repo enterprise/backend combining GitHub PRs and Jira issues. How should individual work items be referenced in the merged queue?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run100

Live evalNeeds ≥20 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+5% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 7 days ago

Other measured skills in the registry, with their headline benchmark lift.