decisiontrail
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
decisiontrail
decisiontrail is a local decision memory for coding agents. When an agent makes a call — which database, which cache, which release cadence — decisiontrail records the decision together with the limits it was made under, so a later session can find the precedent instead of re-deciding from scratch, and can revise it explicitly when the call changes.
Everything is local: decisions live in one JSON file on your machine. There is no account, no login, and no network access. Records can be plain commits or validated decision records (constraints, evidence and an answer); validated records are the ones precedent search draws on. Old decisions are never edited in place — a revision creates a new record and marks the old one superseded, linked in both directions.
This package is a clean-room rebuild of the local surface of the
"elen" MCP server (10 of its 12 tools). The original's two
connected-mode beta tools, elen_incoming and elen_team_search,
need a hosted workspace, a login and an API key, and are not included.
Install
npx @reneturcios/decisiontrail elen_statusOr run the container image:
docker run -i ghcr.io/cloudcorprecords/decisiontrailThe container defaults to the MCP server (see below). Pass CLI arguments to use it as a CLI instead. If you run the container, mount a volume over the store directory if you want decisions to persist.
Related MCP server: horizon-ledger
Where decisions live
Store path: the
DECISIONTRAIL_DBenvironment variable, or~/.decisiontrail/decisions.jsonby default.The file is a single JSON document. A missing file starts empty; writes are atomic (temp file, then rename), so a crash mid-write cannot truncate the store.
DECISIONTRAIL_AGENTandDECISIONTRAIL_PROJECTname the namespace reported byelen_status; both default tolocal.
Decision IDs are sequential: d-1, d-2, … in creation order.
CLI
One bin serves both faces. Each tool is a subcommand taking a single JSON argument (optional for the no-argument tools); results print as JSON.
# Check the namespace before committing
decisiontrail elen_status
# Commit a decision with its limits
decisiontrail elen_commit '{"text":"Use Postgres for sessions","constraints":["must be durable"],"domain":"storage"}'
# Record a validated decision
decisiontrail elen_log_decision '{"constraints":["latency under 100ms"],"evidence":["benchmark run 42"],"answer":"Adopt the in-process cache","domain":"performance"}'
# Before a new decision, ask for pointers; then search validated precedents
decisiontrail elen_suggest '{"query":"session storage"}'
decisiontrail elen_search_precedents '{"query":"session storage"}'
# Read one record in full, revise one, survey the store
decisiontrail elen_expand '{"id":"d-1"}'
decisiontrail elen_supersede '{"id":"d-1","text":"Use Redis for sessions with Postgres fallback"}'
decisiontrail elen_get_context
decisiontrail elen_get_competency
decisiontrail elen_consolidate
decisiontrail --versionExit codes: 0 on success, 1 for a domain error (for example an
unknown decision ID — the error names the ID), 2 for usage errors
and invalid arguments.
MCP server
npx -y @reneturcios/decisiontrail mcpMCP client configuration:
{
"mcpServers": {
"decisiontrail": {
"command": "npx",
"args": ["-y", "@reneturcios/decisiontrail", "mcp"]
}
}
}Tools
Tool | Input | What it returns |
|
| The new unvalidated decision record. |
|
| A validated decision record built from those inputs. |
|
| The full record for that decision ID. |
|
| Short pointers ( |
|
| Full records matching the query, validated records only. |
|
| The new successor record; the old record is marked superseded and the two are linked both ways. |
| — | Active decision threads grouped by domain, with per-domain counts and the latest text, plus totals. |
| — | Agent, project, database path, and counts (total / active / superseded / validated). |
| — | Validated-record counts per domain and the validated share (coverage) of all records, derived from validated history only. |
| — | Duplicate pairs (same domain, token Jaccard at least 0.6), the clusters they form, and stale candidates (active records older than 90 days). Changes nothing. |
How matching works
elen_suggest and elen_search_precedents rank by case-insensitive
token overlap between the query and a record's text plus constraints:
the score is the number of distinct query tokens present. Records with
no overlap are omitted; ties are broken by recency, newest first.
A record counts as validated only when it was created with
elen_log_decision. A committed decision that is later superseded
keeps its history: supersedes / superseded_by link the chain and
the old record's status becomes superseded.
Development
npm install
npm run build # tsc -> dist/
npm test # node test runner over test/*.test.ts, temp dirs, no network
npm run dev # run the CLI from source via tsxLicense
MIT — see LICENSE.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Knowledge accumulation for AI coding agents. Records decisions, problems, and insights as context.
Give your AI agent persistent, governed memory for every project. At task start it recalls the approved decisions, conventions, risks and architecture (semantic search, ranked by importance); at close it proposes what was learned as typed memories that you review and approve — governance, not a notes dump. Agents propose, humans govern: edits go back to pending and deletion is human-only by design. Connect Claude Code, Cursor, Claude Desktop or any MCP client in two minutes with just your API key — hosted (nothing to install) or locally via `uvx solucortex-mcp`. Built by SoluAI and dogfooded daily: SoluCortex is developed using its own living memory.
Memory for coding agents: the decisions, the dead ends, and where the last session stopped.
Query your team's AI coding history and act on HypAware recommendations from any MCP client.
Related MCP Servers
- AlicenseAqualityCmaintenanceEnables coding agents to maintain durable, private, local memory by storing and linking records across collections, with structured review, search, and audit history.23MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI coding agents to read and manage a Git-native decision ledger, providing tools for querying decisions, evidence, cross-repository workspaces, validation, and commit tracing, all operating locally and without a cloud service.MIT
- AlicenseNot gradedqualityCmaintenanceEnables an AI coding agent to retrieve decisions captured from earlier sessions, returning recent decisions and topic-matched ones cited by id, plus expanded context, all from a local SQLite database. It also renders briefs at session start so later work starts from what was already settled.MIT
- AlicenseBqualityCmaintenanceEnables coding agents to persist and retrieve ADR-style decisions and machine-checkable constraints as git-friendly markdown in a repository, linked to requirement ids, tests and commits. Agents can pull relevant context before a task, check a proposed change against accepted constraints before committing, and record, supersede or search decisions over a local stdio connection.740 npmMIT