* feat: move to xml top tags for ebtter llm parsing and structure
- Orchestrator is now purely an orchestrator
- Added new calrify phase for immediate user erequest understanding and task parsing before workflow
- Enforce review/ critic to plan instea dof 3x plan generation retries for better error handling and self-correction
- Add hins to all agents
- Optimize defitons for simplicity/ conciseness while maintaining clarity
* feat(critic): add holistic review and final review enhancements
* Clarify accessibility criteria descriptions
I have not been able to look through everything in the accessibility instruction file. But the changes I'm submitting attempt to clarify / correct content that needed it.
* Potential fix for pull request finding
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
* Apply suggestion from @scottaohara
* Apply suggestion from @scottaohara
---------
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
* feat(skills,agents): add LinkedIn post formatter skill and writer agent
Add self-contained LinkedIn post formatting skill with Unicode bold/italic
character mapping and engagement-optimized post structure patterns.
Add LinkedIn Post Writer agent for transforming raw content into
copy-paste-ready LinkedIn posts. No external service dependencies.
* fix: address Copilot review comments on PR #1392
- Correct Unicode range header to include digit range (U+1D7EC–U+1D7F5)
- Add explicit instruction to load references/unicode-charmap.md
- Fix LinkedIn casing in README.agents.md (Linkedin → LinkedIn)
* fix: use display name for LinkedIn agent to preserve brand casing
The name field now uses 'LinkedIn Post Writer' instead of the slug
format, ensuring the auto-generated README table shows correct casing.
---------
Co-authored-by: Shailesh Mishra <shaileshmishra@example.com>
Update tools lists in tdd-red, tdd-green, tdd-refactor, and github-actions-expert
agents to use the current VS Code Copilot built-in tool names.
Old names were silently ignored by VS Code (documented behaviour: 'If a given
tool is not available when using the custom agent, it is ignored.').
Tool name mapping applied:
- github → github/*
- findTestFiles → search/fileSearch
- runTests → execute/runTests
- runCommands → execute/runInTerminal
- terminalCommand → execute/runInTerminal
- codebase → search/codebase
- filesystem → read/readFile
- problems → read/problems
- testFailure → execute/testFailure
- terminalLastCommand → read/terminalLastCommand
- githubRepo → github/*
Also added missing tools present in the current VS Code tool reference:
- execute/getTerminalOutput
- read/terminalSelection
Additionally, generalised C#-specific content in TDD agents to be polyglot,
covering JavaScript/TypeScript (Jest/Vitest), Python (pytest), Java/Kotlin
(JUnit 5), and C# (.NET) — making the agents useful to the broader community.
Reference: https://code.visualstudio.com/docs/copilot/reference/copilot-vscode-features#_chat-tools
Co-authored-by: Ming <oncwnuAUeYAFR4UGlq8BEJq8Jy-k@git.weixin.qq.com>
* Update modernize-dotnet plugin to 1.0.1026-preview1
* Update modernize-dotnet version in marketplace.json to 1.0.1026-preview1
---------
Co-authored-by: Antonio Villanueva <anvillan@microsoft.com>
* Add whatidid skill — Copilot impact report generator
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Fix: use single quotes for description in SKILL.md frontmatter
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Convert from skill to external plugin per reviewer feedback
Address @aaronpowell's review: convert whatidid from a bundled skill
to an external plugin referencing microsoft/What-I-Did-Copilot.
This improves supply chain transparency and stays current with
the source repo.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* feat: add code-tour skill for AI-generated CodeTour walkthroughs
* fix: trim SKILL.md from 645 to 432 lines (under 500 limit)
Reduce persona table to top 10, condense verbose examples and notes,
trim redundant anti-patterns, compress step type docs and PR recipe.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: run npm run build to update README with code-tour skill
Addresses review feedback from @aaronpowell
* fix: add missing scripts/ and references/ files referenced in SKILL.md
Addresses reviewer feedback — SKILL.md referenced bundled files
(validate_tour.py, generate_from_docs.py, codetour-schema.json,
examples.md) that were not included in the PR.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: run npm run build to update skills README with new assets
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat: add azure-functions-csharp.instructions.md
Added guidelines and best practices for building Azure Functions in C# using the isolated worker model.
* feat: add azure-durable-functions-csharp.instructions.md
Added guidelines and best practices for building Azure Durable Functions in C# using the isolated worker model.
* docs: add config & testing sections to durable functions instructions
Added detailed configuration and testing guidelines for Azure Durable Functions in C#.
* Fix capitalization in applyTo path for local.settings.json
Changed Local.settings.json to local.settings.json in the applyTo directive to match the actual filename used in Azure Functions projects.
* update through npm build
* feat(instructions): update security, a11y, and performance to 2025-2026 standards
Security: OWASP 2025 (55 anti-patterns, AI/LLM section, 6 frameworks)
Accessibility: WCAG 2.2 AA (38 anti-patterns, legal context EAA/ADA, 4 frameworks)
Performance: CWV (50 anti-patterns, Next.js 16, Angular 20, modern APIs)
* fix(instructions): use globalThis.scheduler to prevent ReferenceError
Access scheduler via globalThis to safely handle environments where
the Scheduling API is not declared as a global variable.
* fix(instructions): correct regex patterns and harden SSRF example
- AU1: anchor jwt.verify lookahead inside parentheses
- AU2: anchor jwt.sign lookahead, add expiresIn alternative
- AU7: fix greedy .* before negative lookahead in OAuth state check
- I5: resolve all DNS records, add TOCTOU production note
- K2: add closing delimiters and multi-digit support to tabindex regex
* fix(instructions): enhance SSRF IP validation with IPv4-mapped IPv6
Normalize IPv4-mapped IPv6 addresses (::ffff:127.0.0.1) before
checking private ranges, preventing bypass via mapped addresses.
* fix(instructions): add noscript fallback for deferred CSS pattern
Without JS, the media="print" + onload pattern leaves the stylesheet
inactive. The noscript tag loads it normally when JS is disabled.
* fix(instructions): add execFileSync to I3 command injection detection
The BAD example uses execFileSync but the regex only matched exec,
execSync, and execFile — missing the sync variant.
* fix(instructions): cover full IPv6 link-local range in SSRF check
fe80::/10 spans fe80-febf (fe8*, fe9*, fea*, feb*). Previous regex
only matched fe80::. Also use normalized variable for consistency.
* fix(instructions): adjust SSRF wording and downgrade reduced-motion severity
- SSRF: replace "full DNS/IP validation" with accurate wording that
acknowledges TOCTOU limitation
- V5: downgrade prefers-reduced-motion from IMPORTANT to SUGGESTION,
remove 2.2.2 (A) reference since it's an AAA enhancement
* fix(instructions): rename AU4 heading to include SHA-256
The heading said "Weak Password Hash (MD5/SHA1)" but the detection
regex and BAD example both use SHA-256. Renamed to "Fast Hash for
Passwords" which better describes the actual anti-pattern.
* fix(instructions): clarify WCAG 2.2 SC 4.1.1 status as obsolete
SC 4.1.1 Parsing is still present in the WCAG 2.2 spec but marked
as obsolete (always satisfied). Changed wording from "removed" to
"obsolete" for accuracy.
* fix(instructions): rename I1 example vars to avoid TS redeclaration
Copy-pasting the I1 SQL injection example as a single block failed with a
TypeScript redeclaration error because both BAD and GOOD snippets used
`const result`. Rename to `unsafeResult`/`safeResult` so the block remains
copy-pasteable into a single scope.
* fix(instructions): migrate I3 example to async execFile with bounds
The I3 command injection example used `execFileSync` in both BAD and GOOD
paths, which (a) redeclared `const output` in the same block and (b) blocks
the Node event loop in server handlers, amplifying DoS impact.
Switch the GOOD/BEST paths to a promisified `execFile` call with explicit
`timeout` and `maxBuffer` bounds, and rename variables to
`unsafeOutput`/`safeOutput` so the snippet stays copy-pasteable. Add a
trailing note recommending async child_process APIs for server code.
* fix(instructions): align AU6 heading with session fixation example
The AU6 heading claimed "Session Not Invalidated on Password Change" but
the mitigation example showed `req.session.regenerate`, which is the
canonical defense against session fixation on login rather than bulk
invalidation after a credential change.
Rename the anti-pattern to "Missing Session Regeneration on Login (Session
Fixation)" so it matches the example, and add a trailing note pointing to
the complementary practice of invalidating other active sessions for the
user on password change (e.g., via a `tokenVersion` counter).
* fix(instructions): make L1 critical CSS pattern CSP-compatible
The L1 "GOOD" snippet relied on an inline `onload="this.media='all'"`
handler on a `<link>` tag. Under a strict CSP that disallows
`'unsafe-inline'` / `script-src-attr 'unsafe-inline'`, inline event
handlers are blocked, so the stylesheet would never activate and users
would hit a styling regression.
Replace the pattern with build-time critical CSS extraction
(Critters/Beasties/Next.js `optimizeCss`) plus a normal
`<link rel="preload" as="style">` and standard `<link rel="stylesheet">`.
Add a trailing note explaining why the older inline-onload trick breaks
under strict CSP and how to defer non-critical CSS with an external
script when deferral is truly needed.
The tools field used non-standard names (codebase, terminalCommand,
fetch_webpage) that are not recognized as valid tool aliases. Per the
custom agents configuration docs, unrecognized tool names are silently
ignored, which effectively gave Ember zero tools.
Removing the tools field entirely enables all available tools by default,
which is the correct behavior for a general-purpose AI partner agent.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
LSP servers are only loaded on Copilot CLI startup, so the user
must exit and re-launch for the new configuration to take effect.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Rewrite description to emphasize code intelligence capabilities
(go-to-definition, find-references, hover) so the coding agent
triggers the skill when it needs deeper code understanding, while
still matching explicit LSP setup/configuration requests.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Replace pylsp primary + pyright alternative with a single pyright
recommendation for Python. Go already correctly recommends gopls.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Replace csharp-ls/OmniSharp with roslyn-language-server in the .NET
section, matching the official dotnet/skills reference configuration.
Addresses review feedback from @aaronpowell in PR #1272.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Improve agent skills instructions with quality authoring patterns
Add practical, platform-agnostic authoring guidance to
agent-skills.instructions.md based on community best practices:
- Expand applyTo to cover all skills/**/SKILL.md paths
- Add writing guidance for every recommended body section with examples
(When to Use, Prerequisites, Step-by-Step Workflows, Gotchas,
Troubleshooting, References)
- Add 'Writing High-Impact Skills' section with 5 principles:
- Focus on what the agent doesn't know
- Context budget awareness for descriptions
- Gotchas as highest-signal content
- Flexible guidelines over rigid steps
- Progressive disclosure for large skills
- Fix description constraints: add 10-char minimum, single-quote wrapping
- Reconcile Step-by-Step Workflows with Flexible Guidelines advice
- Consolidate Workflow Execution Pattern under Common Patterns
- Standardize terminology (agent vs AI agent)
- Align line thresholds (200 soft / 500 hard)
- Update validation checklist with new quality criteria
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Update instructions/agent-skills.instructions.md
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* Address PR review: make instructions platform-agnostic
- Replace 'GitHub Copilot'-specific language with generic 'the agent'
- Add skills/<skill-name>/ to Directory Structure table for repo layouts
- Fix relative path in flexible guidelines example (./references/)
- Update description and intro to be platform-neutral
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* Revert terminology changes per reviewer feedback
Restore 'Copilot' and 'AI agent' terminology to match
the original doc's conventions. Keep new substantive content
(Writing High-Impact Skills, Gotchas, expanded examples).
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Aaron Powell <me@aaron-powell.com>
- Adds React 18 and 19 migration orchestration plugins
- Introduces comprehensive upgrade toolkits for migrating legacy React 16/17 and 18 codebases to React 18.3.1 and 19, respectively. Each plugin bundles specialized agents and skills for exhaustive audit, dependency management, class/component API migration, test suite transformation, and batching regression fixes.
- The React 18 toolkit targets class-component-heavy apps, ensures safe lifecycle and context transitions, resolves dependency blockers, and fully automates test migrations including Enzyme removal. The React 19 toolkit addresses breaking changes such as removal of legacy APIs, defaultProps on function components, and forwardRef, while enforcing a gated, memory-resumable migration pipeline.
- Both plugins update documentation, plugin registries, and skill references to support reliable, repeatable enterprise-scale React migrations.
