// Use when mining conversation history during dev work — recovering decisions from the top-level primary session, delegating bulk transcript reading to an explorer, or discovering all sessions tied to a work item across interruptions.
Load into any agent that receives human direction. Use before acting on instructions, recording decisions, or producing artifacts.
Load when establishing shared vocabulary, resolving term conflicts, disambiguating domain terminology, or auditing consistency across a project's vocab. Shared vocabulary is the contract between human intent and agent action — ambiguity is the root failure mode, vocab files are the record.
Load when reading from, writing to, or maintaining the KB. Covers layer model, structural conventions, and operational tooling.
Use whenever you need to delegate work to another agent, run tasks in parallel, coordinate multiple agents, or inspect spawn outputs. Also use when routing work to a specific model or provider.
Load before spawning subagents to protect focus and role fit. Make sure to re-read the available agents before spawning a subagent; choose the most specific owner instead of broad defaults, then write a tight handoff with only the context that subagent needs.
Load when producing any written artifact for humans.
| name | session-mining |
| type | reference |
| description | Use when mining conversation history during dev work — recovering decisions from the top-level primary session, delegating bulk transcript reading to an explorer, or discovering all sessions tied to a work item across interruptions. |
| model-invocable | true |
Design decisions, rejected alternatives, and discovered constraints often live in conversation instead of code or docs. Compaction makes that context expensive to recover later, so mine it before the next implementation or documentation step starts.
Start from the top-level conversation. $MERIDIAN_CHAT_ID points to the primary session at the root of the chat tree, regardless of how deep in the spawn tree you are. That root context holds the primary's decisions and framing — usually the highest-leverage read.
Use a narrow read first, then widen only if needed:
meridian session log "$MERIDIAN_CHAT_ID" --last 20
When the question spans long histories or multiple sessions, spawn @session-explorer for transcript mining and synthesis. That keeps your context window focused on decisions instead of raw transcript paging.
meridian spawn -a session-explorer \
--prompt-file session-mine.md
If you are the @session-explorer, mine directly rather than spawning recursively.
When work spans interruptions, reopenings, or multi-day execution, list all related sessions before mining. Reading only the current session misses prior decisions.
meridian work sessions <work_id> --all
Skip mining if the spawning prompt already includes the needed rationale, or if current design/plan artifacts already capture the decisions you need. Skim artifacts first, and mine only the gaps.