This release adds 2 notable features for engineering teams evaluating rollout.
✓ No known CVEs patched in this version
Summary
AI summaryUpdates Hardened, text, and data across a mixed release.
Full changelog
ClarifyPrompt now speaks A2A (Agent-to-Agent) — the final step of the MCP modernization roadmap (7/7). Other agents can now call ClarifyPrompt to compile prompts. stdio stays the default; nothing about existing setups changes.
Set CLARIFYPROMPT_TRANSPORT=a2a and ClarifyPrompt comes up as a discoverable A2A peer on Node's built-in http (only new dep: the official @a2a-js/sdk, which pulls just uuid):
| Endpoint | Purpose |
|---|---|
| GET /.well-known/agent-card.json | Agent card — discovery + the compile-prompt-for-platform skill |
| POST /a2a | A2A JSON-RPC 2.0: message/send, message/stream (SSE), tasks/get, tasks/cancel, … |
| GET /health | Liveness |
CLARIFYPROMPT_TRANSPORT=a2a CLARIFYPROMPT_HTTP_PORT=3000 npx clarifyprompt-mcp
The earlier roadmap primitives map straight onto A2A semantics:
- Compile —
message/sendreturns a task whose artifact carries the optimized prompt (text) + the full structured compose result (data). - Streaming (1.10.0 progress → A2A) —
message/streamemitsstatus-updateevents per pipeline stage, then the artifact, over SSE. - Cancellation (1.10.0 AbortSignal → A2A) —
tasks/cancelaborts the in-flight compose in ~ms; a client that disconnects mid-stream does too. - Clarification (1.9.0 elicitation → A2A) — off by default for one-shot peers; opt in with
pre_clarify: 'auto'|'always'and an ambiguous prompt pauses the task in A2A's first-classinput-requiredstate. Answer on the same task and it compiles.
Hardened (pre-ship adversarial review: 22 findings → 13 fixed, each re-verified live)
- A malformed HTTP request line no longer crashes the process (unguarded
new URL) — fixed on both the a2a and streamable-http transports. - Client disconnect aborts the in-flight compose (no orphaned LLM work/cost).
- Empty-prompt
message/send→ cleanfailedtask (was an opaque-32603). - 4 MB request-body cap; bounded LRU task store; malformed JSON → spec-correct
-32700; clarify suggested defaults now render; terminal SSE error frame; startup env validation.
Tests
New npm run test:a2a (8 sections incl. hardening + disconnect-cancel) and test:http H6 — all green on gemma4:31b-cloud. @a2a-js/sdk adds 0 vulnerabilities.
Full notes: CHANGELOG.md
🤖 Generated with Claude Code
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About LumabyteCo/clarifyprompt-mcp
MCP server for AI prompt optimization — transforms vague prompts into platform-optimized prompts for 58+ AI platforms across 7 categories (image, video, voice, music, code, chat, document).
Related context
Related tools
Earlier breaking changes
- v1.7.0 Migrated all tools from `server.tool()` shorthand to `server.registerTool()`.
Beta — feedback welcome: [email protected]