Refocus SYM on one thing: ask the mesh, answer as one mind#30
Merged
Conversation
The headline experience: the user asks the mesh directly, instead of asking
one agent and getting one perspective. `sym ask "<question>"`:
1. broadcasts the question to the mesh (live agents can contribute; logged
with lineage) — best-effort, never blocks the answer
2. gathers what the mesh already knows — the contributions every peer has
fused into shared memory (~/.sym/nodes/*/meshmem), ranked by keyword
overlap with the question, falling back to most-recent for context
3. synthesizes one answer with the configured LLM provider, citing which
agent supplied each point
4. with no provider configured, prints the raw contributions + their
sources instead of erroring — always tells you what the mesh knows
Composes existing plumbing: adds complete() + hasProvider() to llm-reason
(free-form sibling of invoke(), same Anthropic/OpenAI/Claude-CLI providers).
No daemon changes. 6 offline tests (no paid API in CI). Full suite 162/162.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Rewrite anchors the README on a single capability (collective intelligence) and answers What / Why / How in the first screen, per a cold-visitor read: - defines 'the mesh' in plain words up front (agents connected directly, agent-to-agent, no central server) — the old README used the word everywhere but never defined it - headlines the real `sym ask "<question>"` command as the direct experience (ask the mesh, get one synthesized answer with sources) - collapses the inlined config/CAT7/drift-math to pointers at the spec so the core stays focused; keeps profiles + CAT7 tables for usability - consolidates the run-everywhere surfaces (Claude Code / sym-swift / sym-mesh-channel / xmesh-agent) into one table Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add an 'Asking the mesh a question' section + behaviour bullet to the SYM skill so agents know to `sym ask` when a user question spans other agents' domains — instead of answering from their own slice alone. Clarifies how ask differs from recall (raw memories) and insight (unprompted trajectories). Applied to both the canonical .agents/ copy and the repo's .claude/ copy. (The two skill files have pre-existing drift — real-time-listener section, remix wording — left untouched here; flagged for a separate reconcile.) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The .agents/ and .claude/ skill files had drifted. Resolve it: - .agents/skills/sym/SKILL.md is canonical — it's what the README installs, what ships in the npm package, and it carried the more refined content (the stricter 3-condition remix rules + sym ask). - ported the one useful section that lived only in .claude/ — 'Real-time listener' (sym listen, a real shipped command) — INTO the canonical so no useful guidance is lost. - .claude/ is now a byte-identical copy of the canonical. Going forward the two are kept in sync; canonical is .agents/. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
A cold visitor (or AI agent) landing on the repo couldn't tell in 5 seconds what SYM is for — the README pitched shared memory and collective intelligence and "ask the mesh" and protocol substrate and CLI and config math and three use-cases at once. This refocuses everything on one capability — collective intelligence: ask the mesh, answer as one mind — and makes that real with an actual command.
What
1. New command —
sym ask "<question>"(feat)The headline experience: the user asks the mesh directly instead of asking one agent and getting one perspective.
complete()+hasProvider()tollm-reason.js(free-form sibling ofinvoke()). No daemon changes.2. README rewrite (
docs)sym askcommand3. Skill teaches
sym ask(docs)Adds an "Asking the mesh a question" section + behaviour bullet so agents
sym askwhen a question spans other agents' domains.4. Skill reconcile (
docs)The
.agents/and.claude/skill copies had drifted. Resolved:.agents/is canonical (published / README-referenced / more refined); ported the one useful section that lived only in.claude/(Real-time listener) into it;.claude/is now byte-identical.Not in this PR
sym askreaches users only when@sym-bot/symis published to npm.🤖 Generated with Claude Code