English | 中文

ARGP — 0-LLM Deterministic Context Compaction for DeepSeek Harness

ARGP (Atomic Reference Graph Pruning) is a third-party CompactionEngine for DeepSeek Harness (dsh) that compresses conversation context without any LLM calls: instead of rewriting history into a summary, it selectively forgets.

• 0 LLM in the compression phase — pure graph rules, deterministic and convergent.
• Selective forgetting, not rewriting — pruned content stays in the append-only session log and is retrievable via a built-in recall_pruned tool.
• Engine-agnostic seam — mounted as a drop-in replacement for compaction-basic through the standard CompactionEngine interface.

Status: research/validation stage. The full pipeline (mount → prune → recall, transaction invariants) is validated on dsh 0.1.0-rc.6 with DeepSeek v4-flash (see Reproduce). Declarative production mounting (P4) is in progress.

Why

Summarizer-based compaction (e.g. compaction-basic) rewrites history with an LLM at compression time: cost scales with context, information is lossy, and compression ratio is not controllable. ARGP takes the opposite route: dependencies are captured structurally while the conversation happens (a small per-turn annotation), and compaction only evicts atoms in reverse topological order of the citation graph — every atom's token count is known, pruning is deterministic, and the degradation chain converges to budget.

Core mechanics

1. Atomization — history is decomposed into atoms (user / assistant / tool-result). dsh's surface has no standalone tool/call nodes, so call blocks live inside assistant atoms.
2. Graph building — deterministic edges (assistant → its tool results, via toolCallId) plus semantic edges from citation prefixes the assistant declares in its output ({"cites": [...]}).
3. Topological pruning — repeatedly evict atoms with in-degree 0, ordered by edge level → effective importance → last-reference round. Citations to a pruned atom unlock it (dynamic effective in-degree, per pass). U (user) atoms and tombstones are never pruned.
4. Closure lifecycle — completed task closures (roots anchored on task-type user atoms) can be evicted whole, with tombstones that feed the recall index.
5. Recall — recall_pruned(seq) retrieves pruned atoms from the log; list_pruned shows the pruned-node index. Budget: ≤3 calls/turn, ≤5% window per call, ≤10% total.
6. Version dedup — exact-duplicate assistant atoms / same-issuer tool results are pruned in pairs (simplified form of the design's θ=0.8 chain dedup).

Design details, invariants, and implementation-vs-design deviations are tracked in docs/.

Repository layout

Path	Content
src/argp-graph-engine.ts	Main engine (graph build, pruning, closure lifecycle, recall/list tools)
src/argp-t1-engine.ts	Earlier single-transaction validation engine
src/recall-engine.ts / src/probe-engine.ts	Recall / probe helpers
test/	Node test suite (argp-graph-engine.test.ts, chain-unlock.test.ts)
spike/	Repro/validation scripts (each node spike/NN-*.ts is self-contained)
docs/	Design (v1.0), migration design, roadmap, experiment records, design↔impl trace

Quick start

npm install
npm run check        # typecheck + local smoke + unit tests

DeepSeek-backed validation requires a dsh API credential (standard dsh credential location) and runs:

npm run smoke:deepseek   # 10a + 10b + 10d single-turn smokes

Reproduce

Key validation runs (all artifacts are local-only; scripts are committed):

Run	Command	What it validates
50-turn t-long (high thinking)	ARGP_DEEPSEEK_THINKING=enabled node spike/06-tlong.ts	L1/L2/L3 invariants, 7/7 anchors, 7/7 needles via recall
Production-scale	ARGP_DEEPSEEK_THINKING=enabled ARGP_WINDOW_TOKENS=100000 ARGP_RETAIN_TOKENS=33000 ARGP_MAX_PASSES=256 node spike/06-tlong.ts	Large-transaction pruning (34–35 atoms per transaction)
Baseline (compaction-basic)	node spike/07-baseline.ts	Same task with the stock summarizer for contrast
Synthetic 0-LLM	npm run spike8a	28-atom single-transaction pruning with zero LLM calls

Experiment results and claims are recorded in docs/experiment-2026-08-16-separated-contract-probe.md; every number carries its artifact path.

Known platform gaps (feedback to dsh)

Developing a non-LLM compaction backend surfaced three extensibility gaps in the compaction seam. Details and repro scripts in docs/dsh-api-feedback-2026-08-17.md:

1. No structured metadata channel on tool/result replacement — placeholders must clone the original message and may only swap content.
2. compaction/prune is outside the transaction invariant state machine — no native event type for algorithmic eviction; third-party engines must borrow summary semantics with pseudo fields.
3. Headless test assembly silently disables the pressure path — mountAgentLoopTestDependencies does not register tokenMeter, and the pre-step catch swallows the error.

License

MIT