dsh-hypercompact
Deterministic, zero-LLM byte-budget context compaction engine for DeepSeek Harness with byte-exact recall tools.
Install
$ dsh plugin --profile web add dsh-hypercompactPlugin Overview & Capabilities
AI-assisted organization based on the public repository snapshot. The content must be grounded in source evidence and does not replace compatibility or security verification.
Key Capabilities
- Byte-budget and token-ratio automatic context compaction without LLM round-trips
- Verbatim retention and pinning of human constraints and prompt instructions
- Single-line compression and grouping for historical tool calls and outputs
- In-place housekeeping and image offloading to defer full compaction cache misses
- Byte-exact recall and log searching tools (`recall`, `recall_search`) and user commands (`/recall`, `/hypercompact`)
- Intra-turn compaction for long autonomous turns exceeding size budgets
- Automatic recovery and single-retry on context overflow, HTTP 413, or body timeout errors
Useful For
- Preventing upload timeouts on slow, rate-limited, or proxied network connections during extended DSH sessions
- Compressing image-heavy and tool-output-heavy agent workflows without losing user instructions
- Enabling agents to inspect and pull original uncompressed tool logs on demand via exact recall
- Managing long multi-turn software development and autonomous agent sessions within strict byte limits
Who It Fits
- DeepSeek Harness users working on long-running coding or multi-step agent sessions
- Developers operating DSH through proxies, CDNs, or bandwidth-constrained network uplinks
- Engineers requiring verifiable, reproducible context preservation without loss of user constraints
Documented Limitations
- Requires DeepSeek Harness versions between 0.1.5-rc.2 and <0.3.0
- Extractive compaction drops assistant reasoning tokens, which are not carried unless recalled
- Must be enabled via a preset (such as the Hypercompact preset) at session initiation; existing active sessions cannot switch compaction engines mid-conversation
- Images attached directly to user messages are not offloaded until their corresponding turn is compacted
DSH Compatibility
Version-specific runtime evidence collected by DSH Plugin. A missing result means we have not tested that combination yet.
Security Signals
Objective signals discovered from package metadata and source inspection. These are not a guarantee that a plugin is safe.
package.json declares 0 runtime, 0 development, 0 peer, and 0 optional dependencies.
package.json declares DSH bundle metadata.
GitHub reports the repository license as MIT.
A root package.json was captured and can be inspected by the registry.
Public GitHub source metadata is available for this registry snapshot.
Source & Registry Notes
Public provenance, Registry classification, and the latest source check for this entry, kept separate from runtime verification.
- Source repository
- mrbeandev/dsh-hypercompact
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin
- Source checked
- fc09a57 · 2026-10-20
This project is independently indexed from public source information. DSH Plugin is not affiliated with DeepSeek or the plugin author. Always check the author repository before installation.
Repository Activity
- GitHub stars
- 2GitHub stars
- Forks
- 0
- Open issues
- 0
- Last commit
- 2026-10-05
- Last release
- No release detected
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
DeepSeek Harness web-search plugin that adds web search, X search, and focused page fetching with structured evidence.
View plugin →Persistent multi-model agent teams with dynamic DAG orchestration, run monitoring, and token insights for DeepSeek Harness.
View plugin →Vision routing and pixel-level tools for text-only DeepSeek Harness agents with built-in free fallback.
View plugin →Session lifecycle manager for DeepSeek Harness with search, restore, idle cleanup, and diagnosis.
View plugin →