Fix Copilot-only coupling in three Agent Skills by MerverliPy · Pull Request #4453 · github/awesome-copilot

MerverliPy · 2026-10-03T04:31:39Z

Three skills in this collection are built on the Agent Skills specification but are coupled to one host agent in ways that make them misbehave elsewhere. Each fix is one or two lines; the reasoning is longer than the change.

Found while vendor-installing these skills into a non-Copilot agent. The three commits are independent and can be taken separately.

1. `dotnet-best-practices`: the `${selection}` prompt variable is not spec

-Your task is to ensure .NET/C# code in ${selection} meets the best practices …
+Your task is to ensure the .NET/C# code in the scope you were asked about meets the best practices …

${selection} is resolved by Copilot's editor selection context. It is not part of the Agent Skills specification, so outside Copilot it is never substituted and the model receives the literal string ${selection}.

This is not cosmetic here: that sentence is the skill's task statement, so the agent is being told to evaluate a scope that does not exist. The replacement carries the same intent with no host dependency.

2. `build-evidence-map`: the description names one agent

-… Use when Copilot must preserve supporting, contradicting, qualifying, and missing evidence …
+… Use when an agent must preserve supporting, contradicting, qualifying, and missing evidence …

A description is the routing text an agent reads to decide whether to load the skill. Naming one vendor's agent there reduces its reach without adding routing information, and reads as a precondition to every other agent.

3. `harness-engineering`: the harness-surface table is Copilot-only

Three changes in the same direction — harness engineering is not Copilot-specific, but this file presents it as though it were.

a. Trigger list opened with "make a repository more reliable for GitHub Copilot or other coding agents". Named agents are now examples rather than the subject.

b. Discovery list gained two Agent Skills specification locations that were missing:

 - `AGENTS.md`
 - `.github/copilot-instructions.md`
 - `.github/instructions/`
+- `.pi/skills/` or `.agents/skills/`

The discovery step exists to find a repository's agent instructions; a repository keeping its skills in either of those locations was invisible to it.

c. Recommended-artifact table said AGENTS.md or .github/copilot-instructions.md, which reads as a choice between equals. It is not one — AGENTS.md is read by 20+ agents, .github/copilot-instructions.md by one:

-| Always-on agent behavior | `AGENTS.md` or `.github/copilot-instructions.md` |
-| File-scoped guidance | `.github/instructions/*.instructions.md` |
+| Always-on agent behavior | `AGENTS.md` (with `.github/copilot-instructions.md` as a Copilot-only mirror) |
+| File-scoped guidance | nested `AGENTS.md`, or `.github/instructions/*.instructions.md` for Copilot |
+| Reusable agent workflows | a skill folder with `SKILL.md` |

The skill row is new. The table had no place for skills despite appearing in a skills collection.

Where the Copilot-only mechanism is genuinely the right answer, it is still named — and still named for Copilot. Nothing was weakened or removed.

Verification

No frontmatter field was added or removed; each skill still validates against the Agent Skills rules (name ^[a-z0-9-]+$ ≤64 chars, folder/name match, description 10–1024 chars), checked with a port of eng/validate-skills.mjs.
No bundled references/ or scripts/ file was touched — git diff --stat shows 3 files, 7 insertions, 5 deletions.
The three skills still route and load; the edits change wording and one variable, not control flow or rules.

The description said "Use when Copilot must preserve supporting, contradicting, qualifying, and missing evidence". A skill description is the routing text an agent reads to decide whether to load the skill; naming one vendor's agent there narrows it without adding information, and reads as a precondition to an agent that is not Copilot. Same sentence, agent-neutral: "Use when an agent must preserve ...".

`${selection}` is a GitHub Copilot prompt variable resolved by the editor's selection context. It is not part of the Agent Skills specification, so outside Copilot it is not substituted at all and the instruction reaches the model as the literal string "${selection}". That is worse than a cosmetic problem for this skill: the sentence is the task statement, and it currently tells the agent to check the best practices of "${selection}" - a scope that does not exist. Rewritten as "the .NET/C# code in the scope you were asked about", which carries the same intent without depending on a host that may or may not be there. Found while porting this skill to a non-Copilot agent; the file is otherwise unchanged.

Three changes, all in the same direction. Harness engineering is not Copilot-specific, but this file presented it as though it were. 1. The trigger list opened with "make a repository more reliable for GitHub Copilot or other coding agents". Named agents now appear as examples rather than as the subject, and `pi` and `Claude Code` are named alongside it. 2. The discovery list now includes `.pi/skills/` and `.agents/skills/`. Both are Agent Skills specification locations, so a repository that keeps its skills there was invisible to the discovery step that exists to find them. 3. The recommended-artifact table said "Always-on agent behavior: `AGENTS.md` or `.github/copilot-instructions.md`", which reads as a choice between equals. It is not: `AGENTS.md` is read by 20+ agents, `.github/copilot-instructions.md` by one. `AGENTS.md` is now listed first with the Copilot file described as a mirror, the file-scoped row offers nested `AGENTS.md` before the Copilot-only path, and a row was added for skills, which the table had no place for despite being the subject it appears in. No rule was weakened: where the Copilot-only mechanism is the right one it is still named, and it is still named for Copilot.

github-actions · 2026-10-03T04:32:18Z

🔒 PR Risk Scan Results

Scanned 3 changed file(s).

Severity	Count
🔴 High	0
🟠 Medium	0
ℹ️ Info	0

✅ No matching risk patterns were detected in changed files.

This is an automated soft-gate report. Findings indicate review targets and do not block merge by themselves.

github-actions · 2026-10-03T04:32:19Z

🚦 Submission status: ⏳ Awaiting automation

Risk tier: merge-risk:low — Documentation, metadata, generated output, or a small update to an existing resource
Required to merge: passing submission-gate checks plus 1 approval from reviewers with write access.

Why this tier

Small update (14 changed lines) to existing resources with no added or removed files

Automated checks

Check	Status	Details
Line endings	✅ Passed	Passed · logs
Spelling	✅ Passed	Passed · logs
Generated README consistency	✅ Passed	Passed · logs
Skill validation	✅ Passed	Passed · logs
Skill lint (vally)	✅ Passed	Passed · logs
Risk scan	✅ Passed	Passed · logs
Contributor reputation	✅ Passed	Passed · logs
Duplicate resource scan	✅ Passed	Passed · logs
PR quality signal	⏭️ Skipped	Skipped by its workflow · logs
Contributor risk signal	🔧 Infrastructure failure	The contributor check succeeded but its result artifact was missing, unreadable, or for another commit · logs

Action needed

🔧 Contributor risk signal hit an automation problem that is not caused by your contribution. Comment /rerun-checks to retry; maintainers are notified if it keeps failing.

Review

Approvals: 0/1
Assigned reviewer: not assigned yet — comment /request-review to ask for one
Review target date: not set
Still needed: 1 more approval(s)
No staffed reviewer pool owns these files yet; any reviewer with write access counts as the resource owner.

Commands

Command	Who	What it does
`/rerun-checks`	PR author, maintainers	Re-runs failed or incomplete checks and re-evaluates this gate
`/request-review`	PR author, maintainers	Asks the review rotation to assign a reviewer (adds `needs-reviewer`)

_{Updated for f5502f3 · gate run · This comment is maintained automatically — see submission gate docs.}

github-actions · 2026-10-03T04:33:46Z

🔍 Vally Lint Results

✅ All checks passed

Scope	Checked
Skills	3
Agents	0
Total	3

Severity	Count
❌ Errors	0
⚠️ Warnings	0
ℹ️ Advisories	0

Summary

Level	Finding
ℹ️	✅ build-evidence-map (2/2 checks passed)
ℹ️	✓ [spec-compliance] All 1 skill(s) are spec-compliant.
ℹ️	✓ spec-compliance: All spec checks passed.
ℹ️	✓ [valid-refs] All file references across 1 skill(s) are valid.
ℹ️	✓ valid-refs: All file references resolve to existing files within the skill directory.
ℹ️	1 skill(s) linted, 1 passed
ℹ️	✅ dotnet-best-practices (2/2 checks passed)
ℹ️	✓ [spec-compliance] All 1 skill(s) are spec-compliant.
ℹ️	✓ spec-compliance: All spec checks passed.
ℹ️	✓ [valid-refs] All file references across 1 skill(s) are valid.

Full linter output

### Linting skills/build-evidence-map
✅ build-evidence-map (2/2 checks passed)
    ✓ [spec-compliance] All 1 skill(s) are spec-compliant.
        ✓ spec-compliance: All spec checks passed.
    ✓ [valid-refs] All file references across 1 skill(s) are valid.
        ✓ valid-refs: All file references resolve to existing files within the skill directory.

1 skill(s) linted, 1 passed

### Linting skills/dotnet-best-practices
✅ dotnet-best-practices (2/2 checks passed)
    ✓ [spec-compliance] All 1 skill(s) are spec-compliant.
        ✓ spec-compliance: All spec checks passed.
    ✓ [valid-refs] All file references across 1 skill(s) are valid.
        ✓ valid-refs: All file references resolve to existing files within the skill directory.

1 skill(s) linted, 1 passed

### Linting skills/harness-engineering
✅ harness-engineering (2/2 checks passed)
    ✓ [spec-compliance] All 1 skill(s) are spec-compliant.
        ✓ spec-compliance: All spec checks passed.
    ✓ [valid-refs] All file references across 1 skill(s) are valid.
        ✓ valid-refs: All file references resolve to existing files within the skill directory.

1 skill(s) linted, 1 passed

Copilot

Copilot review overview

🟡 Changes recommended

The generated skills catalog still contains the old Copilot-specific description.

Review effort: Balanced
Findings: 1

Open (1)

Regenerate skills catalog to update stale Copilot-only description · New

What changed in this PR

Decouples three Agent Skills from Copilot-specific assumptions.

Changes:

Replaces ${selection} with host-neutral scope wording.
Generalizes routing and harness guidance across coding agents.
Adds portable skill discovery and artifact recommendations.

File	Description
`skills/harness-engineering/SKILL.md`	Expands multi-agent guidance.
`skills/dotnet-best-practices/SKILL.md`	Removes a Copilot-specific variable.
`skills/build-evidence-map/SKILL.md`	Generalizes skill routing text.

💡 Configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

 ---
 name: build-evidence-map
-description: 'Build an auditable evidence map for a contested technical choice, research synthesis, proposal review, or consequential decision. Use when Copilot must preserve supporting, contradicting, qualifying, and missing evidence with exact source regions instead of collapsing disagreement into prose.'
+description: 'Build an auditable evidence map for a contested technical choice, research synthesis, proposal review, or consequential decision. Use when an agent must preserve supporting, contradicting, qualifying, and missing evidence with exact source regions instead of collapsing disagreement into prose.'


`docs/README.skills.md` is generated from the skills' frontmatter by `eng/update-readme.mjs`, and `validate-readme` fails any pull request where the generated output does not match the sources. Changing a description without regenerating it leaves the catalogue advertising the old text. Output of `npm start`, unedited: one line, the build-evidence-map row. Nothing else in the generated tree moved, which is the useful part of the check - it confirms the committed generated files were otherwise in sync. Pairs with a3556e9; that commit's description change is what this line now matches.

Copilot

Copilot review overview

🟢 Approval recommended

The focused documentation changes are consistent, portable, and introduce no unresolved issues.

Review effort: Balanced
Findings: 1

Open (1)

Regenerate skills catalog to update stale Copilot-only description

aaronpowell · 2026-10-07T06:13:43Z

Given that the repo is called Awesome Copilot, it makes sense that things are designed to optimise how Copilot behaves.

MerverliPy added 3 commits October 2, 2026 23:31

Copilot AI balanced review requested due to automatic review settings October 3, 2026 04:31

MerverliPy requested a review from a team as a code owner October 3, 2026 04:31

github-actions Bot added skills PR touches skills merge-risk:low requires-submitter-fixes Submission has quality-gate findings that submitter must fix before maintainer review labels Oct 3, 2026

Copilot started reviewing on behalf of MerverliPy October 3, 2026 04:32

View session

Copilot AI reviewed Oct 3, 2026

View reviewed changes

Copilot AI balanced review requested due to automatic review settings October 3, 2026 04:37

MerverliPy requested a review from a team as a code owner October 3, 2026 04:37

Copilot started reviewing on behalf of MerverliPy October 3, 2026 04:37

View session

github-actions Bot added awaiting-automation and removed requires-submitter-fixes Submission has quality-gate findings that submitter must fix before maintainer review labels Oct 3, 2026

Copilot AI reviewed Oct 3, 2026

View reviewed changes

github-actions Bot added ready-for-review Submission passed intake validation and is ready for maintainer review awaiting-automation and removed awaiting-automation ready-for-review Submission passed intake validation and is ready for maintainer review labels Oct 3, 2026

aaronpowell closed this Oct 7, 2026

Navigation Menu

Repository navigation

Fix Copilot-only coupling in three Agent Skills - #4453

Fix Copilot-only coupling in three Agent Skills#4453
MerverliPy wants to merge 4 commits into
github:mainfrom
MerverliPy:fix/skills-copilot-only-coupling

MerverliPy commented Oct 3, 2026

Uh oh!

github-actions Bot commented Oct 3, 2026

Uh oh!

github-actions Bot commented Oct 3, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 3, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI left a comment

Uh oh!

aaronpowell commented Oct 7, 2026

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

MerverliPy commented Oct 3, 2026

1. dotnet-best-practices: the ${selection} prompt variable is not spec

2. build-evidence-map: the description names one agent

3. harness-engineering: the harness-surface table is Copilot-only

Verification

Uh oh!

github-actions Bot commented Oct 3, 2026

🔒 PR Risk Scan Results

Uh oh!

github-actions Bot commented Oct 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🚦 Submission status: ⏳ Awaiting automation

Automated checks

Action needed

Review

Commands

Uh oh!

github-actions Bot commented Oct 3, 2026

🔍 Vally Lint Results

Summary

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot review overview

🟡 Changes recommended

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot review overview

🟢 Approval recommended

Uh oh!

aaronpowell commented Oct 7, 2026

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

1. `dotnet-best-practices`: the `${selection}` prompt variable is not spec

2. `build-evidence-map`: the description names one agent

3. `harness-engineering`: the harness-surface table is Copilot-only

github-actions Bot commented Oct 3, 2026 •

edited

Loading