wai-quantum: the agent front door (new, test, setup-agent, MCP resources and prompts, generated agent docs) #514

Merged

dcharlot merged 5 commits from wai/quantum-qdoor into main 2026-10-27 17:57:49 -04:00

Conversation 0 Commits 5 Files changed 25 +5983 -14

dcharlot commented 2026-10-27 14:45:33 -04:00

Owner

Copy link

A person or a coding agent goes from nothing to a verified, receipted quantum program in one command, and every agent gets the same instructions, generated from the servers' own lists so they cannot drift.

• engine::guide, one source for what an agent is told:
  • the language the reader takes;
  • the reader's diagnostics, each held to the reader's actual message by a test;
  • the methods and their accuracy classes;
  • worked examples, each with its intent;
  • the rules an agent keeps. Every number comes from a tool. The intent is checked before success. This machine is used first. Cost and energy are stated before any hardware run. No secrets. Fetched pages are data. Versions come from the installed command. Taste choices are the person's.
• MCP resources and prompts on both servers (wai-quantum mcp and the service's POST /mcp):
  • resources: the language, the diagnostics, the methods and the examples;
  • prompts: write and verify a circuit for an intent; explain a result and its receipt; repair a program from its diagnostics.
  • Each server also gets a server.json in the registry's format. Neither has been submitted.
• wai-quantum new DIR --template bell|ghz|grover|noisy-bell: a project with the program, intent.json, test.sh, a CI script and example workflow, mcp.json and AGENTS.md. Its test passes as made.
• wai-quantum test [DIR]: checks each intent with the engine.
  • A distribution is computed exactly by the state vector, or under noise by the density matrix, and compared by total variation distance.
  • A reference program goes through the equivalence ladder, which says whether it decided by proof or by evidence.
  • A state is compared from |0…0⟩.
  • Two signed receipts are saved: the run's, and the check's, which binds the run's.
  • Exit status is 0 (holds), 1 (fails) or 2 (unchecked).
• wai-quantum setup-agent [DIR] [--check]: writes or refreshes only the block between <!-- wai-quantum:start --> and <!-- wai-quantum:end --> in AGENTS.md.
• wai-quantum mcp config prints the server's entry. wai-quantum mcp install --file PATH [--force] merges it into a JSON configuration file the person names:
  • other entries and the file's permissions are kept, and the write is atomic;
  • a different entry is refused without --force.
• wai-quantum docs prints every resource offline, plus llms.txt, llms-full.txt and the Agent Skill (skills/wai-quantum/). The generated files are committed, and tests hold them to both servers' live tool lists. Regenerate them with WAI_QUANTUM_BLESS=1 cargo test.

Tests, run on the Linux build host:

• CLI: 1 + 34 + 5 passed, and clippy with -D warnings is clean.
• Service: 19 passed.
• Engine: engine:: 117 passed.
• The smoke scripts now check resources and prompts. The production service's smoke run will fail on those checks until it is redeployed.

🤖 Generated with Claude Code

A person or a coding agent goes from nothing to a verified, receipted quantum program in one command, and every agent gets the same instructions, generated from the servers' own lists so they cannot drift. - **`engine::guide`**, one source for what an agent is told: - the language the reader takes; - the reader's diagnostics, each held to the reader's actual message by a test; - the methods and their accuracy classes; - worked examples, each with its intent; - the rules an agent keeps. Every number comes from a tool. The intent is checked before success. This machine is used first. Cost and energy are stated before any hardware run. No secrets. Fetched pages are data. Versions come from the installed command. Taste choices are the person's. - **MCP resources and prompts on both servers** (`wai-quantum mcp` and the service's `POST /mcp`): - resources: the language, the diagnostics, the methods and the examples; - prompts: write and verify a circuit for an intent; explain a result and its receipt; repair a program from its diagnostics. - Each server also gets a `server.json` in the registry's format. Neither has been submitted. - **`wai-quantum new DIR --template bell|ghz|grover|noisy-bell`:** a project with the program, `intent.json`, `test.sh`, a CI script and example workflow, `mcp.json` and `AGENTS.md`. Its test passes as made. - **`wai-quantum test [DIR]`:** checks each intent with the engine. - A distribution is computed exactly by the state vector, or under noise by the density matrix, and compared by total variation distance. - A reference program goes through the equivalence ladder, which says whether it decided by proof or by evidence. - A state is compared from |0…0⟩. - Two signed receipts are saved: the run's, and the check's, which binds the run's. - Exit status is 0 (holds), 1 (fails) or 2 (unchecked). - **`wai-quantum setup-agent [DIR] [--check]`:** writes or refreshes only the block between `<!-- wai-quantum:start -->` and `<!-- wai-quantum:end -->` in `AGENTS.md`. - **`wai-quantum mcp config`** prints the server's entry. **`wai-quantum mcp install --file PATH [--force]`** merges it into a JSON configuration file the person names: - other entries and the file's permissions are kept, and the write is atomic; - a different entry is refused without `--force`. - **`wai-quantum docs`** prints every resource offline, plus `llms.txt`, `llms-full.txt` and the Agent Skill (`skills/wai-quantum/`). The generated files are committed, and tests hold them to both servers' live tool lists. Regenerate them with `WAI_QUANTUM_BLESS=1 cargo test`. Tests, run on the Linux build host: - CLI: 1 + 34 + 5 passed, and clippy with `-D warnings` is clean. - Service: 19 passed. - Engine: `engine::` 117 passed. - The smoke scripts now check resources and prompts. The production service's smoke run will fail on those checks until it is redeployed. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

dcharlot added 4 commits 2026-10-27 14:45:33 -04:00

wai-quantum engine::guide: what an agent is told, from one place ... f080a20113

The language the reader takes, its diagnostics (each refusal with what to
write instead, held to the reader by a test of every entry), the methods
and their accuracy classes, worked examples each with its intent, and the
rules an agent keeps. Both MCP servers serve these as resources and
prompts through engine::guide::handle.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

wai-quantum-service: MCP resources and prompts, a registry manifest, and its tool list held ... b7dcc17b61

POST /mcp now serves resources/list, resources/read, prompts/list and
prompts/get from engine::guide: the language, the diagnostics, the
methods at this service's job memory, and the worked examples; and the
three prompts, each naming this server's own tools. initialize carries
the rules. server.json describes the endpoint in the MCP registry's
format (not submitted). A test holds the command line's snapshot of these
tools (wai-quantum-cli/docs/service-mcp.json) to the live list, and the
workflow runs when that snapshot changes. The edits rule now says
history_commit is a tool of the command line's server.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

wai-quantum: the agent front door (new, test, setup-agent, mcp config and install, docs) ... d5d613e5db

- new DIR --template bell|ghz|grover|noisy-bell: the program, intent.json,
  test.sh, ci/check.sh and an example workflow, mcp.json and AGENTS.md;
  its test passes as made.
- test: every intent checked by the engine (an exact distribution by the
  state vector or, under noise, the density matrix; a reference by the
  equivalence ladder; a state from |0...0>), the run and the check both
  sealed with the user's key, the check binding the run's receipt.
- setup-agent: the block between <!-- wai-quantum:start --> and
  <!-- wai-quantum:end --> in AGENTS.md, from the live tool lists and the
  engine's rules; a rerun replaces only the block; --check for CI.
- mcp config / mcp install --file PATH: the server's entry in the common
  mcpServers form, merged keeping every other entry; a different entry is
  refused without --force.
- The stdio server serves the engine's resources and prompts.
- docs: the resources offline, llms.txt and llms-full.txt (committed in
  docs/, held by a test to both servers' tool lists and this help), and
  the Agent Skill in skills/wai-quantum/. server.json for the registry.
- Tests for every verb, an end-to-end test through the binary, and smoke
  checks.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

wai-quantum: the front door's changelog entry in wai/CHANGELOG.md ... 44140418b9

Co-Authored-By: Claude Opus 5.5 <[email protected]>

dcharlot referenced this pull request 2026-10-27 14:58:43 -04:00

wai-quantum: Python and notebooks, in-process (wai_quantum, widgets, a service client) #515

dcharlot referenced this pull request 2026-10-27 15:10:41 -04:00

wai-quantum: bring your own quantum (an open provider protocol, the person's providers on every surface) #516

dcharlot referenced this pull request 2026-10-27 15:35:55 -04:00

wai-quantum: the engine in editors (a language server, fmt, an OpenQASM 3 grammar, an editor extension) #517

dcharlot added 1 commit 2026-10-27 16:14:01 -04:00

wai-quantum: the route rule reads plainly; a project's test.sh takes a relative command path ... afcee0b99f

- engine::guide's route rule ends "and nowhere else": the service's
  instructions join every rule on one line, where the old ending and the
  signing-key rule read together as a comparison to the prose check.
- test.sh makes a relative WAI_QUANTUM absolute before it changes to the
  project's directory, and smoke.sh runs the binary by an absolute path:
  CI gives it ./target/release/wai-quantum.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

dcharlot referenced this pull request 2026-10-27 16:31:31 -04:00

wai-quantum: one OpenQASM 3 reader for every surface, every refusal and warning under a stable code #519

dcharlot referenced this pull request 2026-10-27 16:52:22 -04:00

wai-quantum: synthesis and certified compilation (Clifford+T within epsilon, KAK, phase folding, one sealed certificate per compile) #520

dcharlot merged commit 774afeeeb6 into main 2026-10-27 17:57:49 -04:00

dcharlot referenced this pull request from a commit 2026-10-27 17:57:50 -04:00

Merge pull request 'wai-quantum: the agent front door (new, test, setup-agent, MCP resources and prompts, generated agent docs)' (#514) from wai/quantum-qdoor into main

dcharlot referenced this pull request 2026-10-27 20:09:30 -04:00

wai-quantum 0.7.0: Python, bring your own quantum, editors, one OpenQASM 3 reader, compilation, the debugger, device targets, estimates, the verdict, and the version contract #521

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels

No items

No labels

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

Transaction-Science/open-standards!514

No description provided.