Repository navigation
feat(adr-001): coherenceScore MCP tool — wire-callable feasibility check - #53
Merged
Merged
Conversation
ADR-001 item #3 was reachable from Rust via coherence::coherence_score but not from the MCP wire. Wire-callable agents had no way to check "is this matrix solvable?" before invoking a solver — they'd just have to call solve and hope. Adds coherenceScore as a pure-TS MCP tool. Mirrors Rust's semantics: margin(i) = (|A[i,i]| - Σ_{j ≠ i} |A[i,j]|) / |A[i,i]| coherence(A) = min_i margin(i) > 0 strictly DD, Neumann convergence guaranteed = 0 boundary case < 0 not DD, iterative solvers may diverge -Inf zero-diagonal row (worst-case incoherence) Lands in src/mcp/server.ts: - New `coherenceScore` tool with x-complexity = Linear - Input schema: just `matrix` (dense or sparse-COO) - Dispatch case + handleCoherenceScore method - Returns {coherence, worst_row, is_strict_dd, note} Composition: completes the predict → check → budget → solve → audit wire pipeline. Now an agent can: 1. estimateComplexityClass(method) — predict class 2. coherenceScore(matrix) — check feasibility (THIS PR) 3. enforceComplexityBudget(...) — implicit via max_complexity_class 4. solve(...) — run the solver 5. verifySparseSolution(...) — audit output (PR #52) All wire-callable, no Rust required. Co-Authored-By: claude-flow <ruv@ruv.net>
1 of 2 tasks
ruvnet
added a commit
that referenced
this pull request
May 19, 2026
…trator The SubLinear orchestrator shipped in Rust (PR #29) had MCP preview primitives (coherenceScore, closureIndices, estimateComplexityClass) and a wire-callable witness (PR #52). But the orchestrator itself wasn't wire-callable — agents had to fall back to `solve` (which returns the full n-vector) instead of the closure-restricted entries the SubLinear path produces. This closes the gap with a pure-TS handler that chains: 1. closure_indices(matrix, delta.indices, closure_depth) 2. for each closure entry: truncated Neumann iteration restricted to the closure (mirrors src/entry.rs math) 3. return Vec<{row, value}> Lands in src/mcp/server.ts: - New `solveOnChangeSublinear` tool with x-complexity = SubLinear - Input schema: matrix, vector (b_new), delta_indices: number[], closure_depth (default 4), max_terms (default 32), tolerance (default 1e-8) - Dispatch case + handleSolveOnChangeSublinear method - Returns {entries: Array<{row, value}>, closure_size, max_terms, closure_depth, note} - Supports both dense and sparse-COO matrix formats Matches Rust solve_on_change_sublinear semantics exactly: - Early-exit when |delta_k[target]| < tolerance - Zero-diagonal in closure → InvalidParams - Empty closure → empty entries - note field warns when closure covers full matrix This completes the wire surface for the change-driven inner loop. Agents can now run the full pipeline over MCP: 1. coherenceScore (PR #53) — is the matrix solvable? 2. closureIndices (PR #54) — how wide is the work? 3. estimateComplexityClass (PRs #30, #42) — what class? 4. solveOnChangeSublinear (THIS PR) — actual SubLinear solve 5. verifySparseSolution (PR #52) — audit output All 5 tools are pure-TS. No Rust/WASM bridge required for any of them. Co-Authored-By: claude-flow <ruv@ruv.net>
1 of 2 tasks
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.
Summary
ADR-001 item #3 was reachable from Rust via `coherence::coherence_score` but not from the MCP wire. Wire-callable agents had no way to check "is this matrix solvable?" before invoking a solver — they'd just have to call solve and hope.
Adds `coherenceScore` as a pure-TS MCP tool. Mirrors Rust's semantics:
```
margin(i) = (|A[i,i]| - Σ_{j ≠ i} |A[i,j]|) / |A[i,i]|
coherence(A) = min_i margin(i)
```
> 0= 0< 0-InfPipeline completion
Now an agent can fully exercise the SubLinear pipeline over MCP:
Test plan
🤖 Generated with claude-flow