Install, launch, and validate Context Lattice from terminal in minutes.

Clone + bootstrap

git clone https://github.com/sheawinkler/ContextLattice.git
cd ContextLattice
BOOTSTRAP=1 scripts/first_run.sh

Guided quickstart

gmake quickstart

`quickstart` now shows resource guidance and lets users choose the best profile for their machine.

Health and dashboard checks

gmake monitor-open
# or terminal-only checks
gmake monitor-check

Verify memory path end-to-end

ORCH_KEY="$(awk -F= '/^CONTEXTLATTICE_ORCHESTRATOR_API_KEY=/{print substr($0,index($0,"=")+1)}' .env)"

curl -fsS -H "content-type: application/json" -H "x-api-key: ${ORCH_KEY}" \
  -d '{"projectName":"_global","fileName":"smoke/cli_check.md","content":"cli smoke ok"}' \
  http://127.0.0.1:8075/memory/write | jq

curl -fsS -H "content-type: application/json" -H "x-api-key: ${ORCH_KEY}" \
  -d '{"query":"cli smoke ok","limit":3}' \
  http://127.0.0.1:8075/memory/search | jq

Installed wrappers for agents/operators

# prescribed task lifecycle
contextlattice doctor --pretty
contextlattice context "what should this agent know before solving the task?" --project contextlattice --pretty
contextlattice resume --project contextlattice --pretty
contextlattice remember "checkpoint summary" --project contextlattice --pretty
contextlattice correct "retrieval was useful" --category useful --project contextlattice --pretty
contextlattice finish "verified result" --success --project contextlattice --pretty

# advanced integration and operator surfaces
contextlattice_search -h
contextlattice_write -h
contextlattice_adopt status --pretty
contextlattice_adopt integrate --repo . --agents codex,claude-code,opencode,hermes-agent,hermes-ultra,omp,mercury-agent,pi,droid --pretty
contextlattice_adopt integrate --repo . --agents codex,claude-code,opencode,hermes-agent,hermes-ultra,omp,mercury-agent,pi,droid --check --pretty
contextlattice_doctor --repo . --agents codex --skip-provider-smoke --pretty
contextlattice_doctor --agents claude-code --skip-provider-smoke --pretty
contextlattice_doctor --agents opencode --skip-provider-smoke --pretty
contextlattice_adopt proof --agents codex --skip-provider-smoke --pretty
contextlattice_agent_adapter profiles
contextlattice_agent_session runtime --pretty
contextlattice_async_inbox_drain --session-id <session-id>
contextlattice_async_inbox_hook --session-id <session-id>
contextlattice_agent_trace --session-id <session-id> --tree
contextlattice_run_advisor "what should this agent know before solving the task?" --pretty
contextlattice_agent_runtime_doctor --pretty
contextlattice_memory_topology --pretty
contextlattice_pack "what should this agent know before solving the task?" --project contextlattice --pretty
contextlattice_retrieval_plan "what evidence does this task owe?" --project contextlattice --pretty

contextlattice_claim_write --project contextlattice --subject release --predicate current_version --object 3.17.2 --pretty

contextlattice_claim_query "what is current?" --project contextlattice --include-superseded --pretty
contextlattice_synthesis_pack_v2 "what can the evidence prove?" --project contextlattice --pretty
contextlattice_policy_candidate --project contextlattice --pretty
contextlattice_policy_status --pretty
contextlattice_skill_draft --payload-file workflow-runs.json --pretty
contextlattice_skill_evaluate --draft-id <draft-id> --payload-file holdouts.json --pretty
contextlattice_skill_export --draft-id <draft-id> --human-approved --approver <identity> --pretty
contextlattice_skill_retire --draft-id <draft-id> --operator <identity> --reason "temporary proof completed" --pretty
contextlattice_skill_foundry_status --pretty
contextlattice_memory_graph_repair --project contextlattice --pretty
contextlattice_memory_graph_efficacy --refresh-cases --project contextlattice --pretty
contextlattice_passport_export "portable task context" --project contextlattice --output passport.json --pretty
contextlattice_passport_verify --file passport.json --pretty
contextlattice_passport_status --pretty
contextlattice_mesh_identity --pretty
contextlattice_mesh_status --pretty
contextlattice_agent_runtime_proof --pretty
contextlattice_agent_adoption_proof --skip-provider-smoke --progress --pretty
contextlattice_skills_index search "browser automation" --pretty
contextlattice_agent_start -h
contextlattice_checkpoint -h
contextlattice_agent_start --soft --compact
contextlattice_search --query "release readiness" --project contextlattice
printf '%s\n' "checkpoint saved" | contextlattice_checkpoint --project contextlattice --topic-path runbooks/setup --file notes/checkpoint.md --stdin

contextlattice_skills_index searches active configured skill roots such as $CODEX_HOME/skills; quarantined/vendor discovery stays separate.

Carry the proof. Leave the machine behind.

# compile bounded evidence into a signed, expiring artifact
contextlattice_passport_export \
  "prepare the release handoff" \
  --project contextlattice \
  --output passport.json \
  --pretty

# verify, diff, or render a replay plan without executing imported text
contextlattice_passport_verify --file passport.json --pretty
contextlattice_passport_replay --file passport.json --pretty

# on the recipient, print only its public age identity
contextlattice_mesh_identity --pretty

# on the sender, grant one recipient one project
contextlattice_mesh_grant create \
  --recipient-id workstation-b \
  --recipient age1... \
  --projects contextlattice \
  --pretty

# encrypt to explicit grants; move the JSON through your own channel
contextlattice_mesh_export \
  --passport-id <passport-id> \
  --grant-ids <grant-id> \
  --output envelope.json \
  --pretty

# receiving import is dry-run first
contextlattice_mesh_import --file envelope.json --pretty
contextlattice_mesh_import --file envelope.json --apply --pretty

Passports are Ed25519-signed. Mesh uses the stable age v1 format with native X25519 recipients and ChaCha20-Poly1305 payload protection. Conflicts become preserved branches, never last-write-wins. ContextLattice performs no delivery network calls and adds no MCP tool surface.

Build the next prompt packet from memory

# compact agent_packet.v1 is the default
contextlattice context "ship release notes without missing prior decisions" --project contextlattice --pretty

# full contracts are available for advanced inspection
contextlattice_synthesis_pack_v2 "prove release readiness" --project contextlattice --full --pretty
contextlattice_pack "ship release notes without missing prior decisions" --project contextlattice --full --pretty
contextlattice_synthesis_pack "ship release notes without missing prior decisions" --project contextlattice --pretty
contextlattice_retrieval_plan "ship release notes without missing proof" --project contextlattice --pretty
contextlattice_claim_query "current release constraints" --project contextlattice --include-superseded --pretty
contextlattice_policy_candidate --project contextlattice --pretty
contextlattice_skill_foundry_status --pretty

contextlattice context returns a bounded proof-carrying packet with evidence, uncertainty, next actions, and an explicit act/verify/abstain gate. Full compiler contracts remain available for harness authors and debugging. The retrieval planner is advisor-only and never silently activates policy.

Give every agent the same memory spine

# one-command live proof
contextlattice doctor --pretty
contextlattice context "ship the current task" --project contextlattice --pretty
contextlattice remember "checkpoint summary" --project contextlattice --pretty
contextlattice resume --project contextlattice --pretty
contextlattice correct "retrieval was stale" --category stale --project contextlattice --pretty
contextlattice finish "verified result" --success --project contextlattice --pretty

# cross-profile adoption proof remains available to operators
contextlattice_adopt proof --agents codex,claude-code,opencode,hermes-agent,hermes-ultra,omp,mercury-agent,pi,droid --skip-provider-smoke --pretty
contextlattice_agent_adoption_proof --skip-provider-smoke --progress --pretty

Use profiles such as codex, claude-code, opencode, hermes-agent, hermes-ultra, omp, mercury-agent, pi, droid, chatgpt-web, and claude-web to keep handoffs consistent across runtimes. The adoption proof reports which skills, context package, session events, graph touches, and handoffs shaped each profile run; the async inbox drain surfaces unseen continuation progress/ready/terminal-degraded steering after normal tool boundaries, with running or pending work labeled warming rather than degraded; runtimes with post-tool hooks can bind the async inbox hook for the same fail-open path; the trace command turns the same evidence into a terminal view or exportable run card; Run Advisor turns that evidence into prompt readiness and next-action guidance.

Review stale audit sessions before cleanup

# dry-run: list only known stale objective-runtime audit/preflight sessions
scripts/agent/contextlattice-session sweep-stale-audits --all-projects --pretty

# confirmed cleanup after review
scripts/agent/contextlattice-session sweep-stale-audits --all-projects --confirm --pretty

Normal users do not need this during everyday memory work. It is an operator safety valve for stale audit/preflight sessions, and dry-run is the default.