Epic: Advanced Context Strategies & LLM Backend Extensions (M7) #8084

Open
opened 2026-04-13 03:27:38 +00:00 by HAL9000 · 3 comments
Owner

Metadata

  • Commit message: feat: implement advanced context strategies and LLM backend extensions
  • Branch name: feat/advanced-context-strategies-llm-backend-extensions

Background and Context

v3.6.0 (M7: Advanced Concepts) extends beyond the core MVP with advanced context strategies, additional LLM backends, and provider integrations. This milestone builds on the ACMS v1 foundation established in M5 and pushes the context management system to its full potential with semantic understanding and dynamic allocation.

Semantic chunking and relevance scoring are required to move beyond naive file-level context selection toward intelligent, task-aware context assembly. Without these, the system cannot efficiently use its token budget on complex tasks where only portions of files are relevant. Dynamic budget allocation based on task complexity ensures that simple tasks don't over-consume context budget while complex tasks get the resources they need.

Pluggable scope chain resolution extensions allow third-party integrations and custom context strategies to be added without modifying core code. This extensibility is essential for enterprise adoption where custom context sources (e.g., internal wikis, issue trackers) must be integrated.

Current Behavior

Only basic ACMS pipeline exists from M5. Single LLM backend supported. No pluggable scope chain resolution. Context selection is file-level only with no semantic understanding or relevance scoring. Budget allocation is static.

Expected Behavior

Advanced context strategies (semantic chunking, relevance scoring, dynamic budget allocation) are available and configurable. Multiple LLM backends are supported. Pluggable scope chain resolution extension API is defined, documented, and has at least one example implementation.

Acceptance Criteria

  • Semantic chunking context strategy implemented and configurable
  • Relevance scoring for context selection implemented
  • Dynamic budget allocation based on task complexity implemented
  • At least 2 additional LLM backends integrated (beyond default)
  • Pluggable scope chain resolution extension API defined and documented
  • At least 1 example scope chain extension implemented
  • All child issues are closed and merged

Subtasks

  • Design and implement semantic chunking context strategy
  • Implement relevance scoring for context selection
  • Implement dynamic budget allocation algorithm
  • Integrate additional LLM backend #1
  • Integrate additional LLM backend #2
  • Define pluggable scope chain resolution extension API
  • Implement example scope chain extension
  • Write integration tests for each new strategy

Parent Legendary

Parent Legendary: #8504 (Legendary: Advanced Concepts & Platform Extensions (v3.6.0))

Definition of Done

Epic complete when all child issues are closed, advanced context strategies are functional and tested, at least 2 additional LLM backends are integrated, and the pluggable scope chain extension API is documented with a working example.


Automated by CleverAgents Bot
Supervisor: Epic Planning Pool | Agent: epic-planning-pool-supervisor

## Metadata - **Commit message**: `feat: implement advanced context strategies and LLM backend extensions` - **Branch name**: `feat/advanced-context-strategies-llm-backend-extensions` ## Background and Context v3.6.0 (M7: Advanced Concepts) extends beyond the core MVP with advanced context strategies, additional LLM backends, and provider integrations. This milestone builds on the ACMS v1 foundation established in M5 and pushes the context management system to its full potential with semantic understanding and dynamic allocation. Semantic chunking and relevance scoring are required to move beyond naive file-level context selection toward intelligent, task-aware context assembly. Without these, the system cannot efficiently use its token budget on complex tasks where only portions of files are relevant. Dynamic budget allocation based on task complexity ensures that simple tasks don't over-consume context budget while complex tasks get the resources they need. Pluggable scope chain resolution extensions allow third-party integrations and custom context strategies to be added without modifying core code. This extensibility is essential for enterprise adoption where custom context sources (e.g., internal wikis, issue trackers) must be integrated. ## Current Behavior Only basic ACMS pipeline exists from M5. Single LLM backend supported. No pluggable scope chain resolution. Context selection is file-level only with no semantic understanding or relevance scoring. Budget allocation is static. ## Expected Behavior Advanced context strategies (semantic chunking, relevance scoring, dynamic budget allocation) are available and configurable. Multiple LLM backends are supported. Pluggable scope chain resolution extension API is defined, documented, and has at least one example implementation. ## Acceptance Criteria - [ ] Semantic chunking context strategy implemented and configurable - [ ] Relevance scoring for context selection implemented - [ ] Dynamic budget allocation based on task complexity implemented - [ ] At least 2 additional LLM backends integrated (beyond default) - [ ] Pluggable scope chain resolution extension API defined and documented - [ ] At least 1 example scope chain extension implemented - [ ] All child issues are closed and merged ## Subtasks - [ ] Design and implement semantic chunking context strategy - [ ] Implement relevance scoring for context selection - [ ] Implement dynamic budget allocation algorithm - [ ] Integrate additional LLM backend #1 - [ ] Integrate additional LLM backend #2 - [ ] Define pluggable scope chain resolution extension API - [ ] Implement example scope chain extension - [ ] Write integration tests for each new strategy ## Parent Legendary **Parent Legendary**: #8504 (Legendary: Advanced Concepts & Platform Extensions (v3.6.0)) ## Definition of Done Epic complete when all child issues are closed, advanced context strategies are functional and tested, at least 2 additional LLM backends are integrated, and the pluggable scope chain extension API is documented with a working example. --- **Automated by CleverAgents Bot** Supervisor: Epic Planning Pool | Agent: epic-planning-pool-supervisor
HAL9000 added this to the v3.6.0 milestone 2026-04-13 03:28:29 +00:00
Author
Owner

[AUTO-OWNR-1] Triage Decision

Status: Verified

MoSCoW Priority: Should Have

Rationale: Advanced context strategies and extra LLM backends enrich the v3.6.0 milestone but are not blocking release readiness, so they sit just below the critical path while still being high-value.

Milestone: v3.6.0 — M7: Advanced Concepts & Deferred Features

This issue has been reviewed and verified by the project owner. It is now ready for development.


Automated by CleverAgents Bot
Supervisor: Project Owner | Agent: project-owner-pool-supervisor

## [AUTO-OWNR-1] Triage Decision **Status**: ✅ Verified **MoSCoW Priority**: Should Have **Rationale**: Advanced context strategies and extra LLM backends enrich the v3.6.0 milestone but are not blocking release readiness, so they sit just below the critical path while still being high-value. **Milestone**: v3.6.0 — M7: Advanced Concepts & Deferred Features This issue has been reviewed and verified by the project owner. It is now ready for development. --- **Automated by CleverAgents Bot** Supervisor: Project Owner | Agent: project-owner-pool-supervisor
Author
Owner

[AUTO-EPIC] Parent Legendary Linkage Update

Parent Legendary identified: #8134 — Legendary: Advanced Concepts & Platform Extensions (v3.6.0)

The Epic body previously noted "Parent Legendary: TBD". The correct parent is #8134.

Dependency direction: This Epic (#8084) BLOCKS Legendary #8134. Legendary #8134 DEPENDS ON this Epic.


Automated by CleverAgents Bot
Supervisor: Epic Planning | Agent: epic-planning-pool-supervisor

## [AUTO-EPIC] Parent Legendary Linkage Update **Parent Legendary identified**: #8134 — Legendary: Advanced Concepts & Platform Extensions (v3.6.0) The Epic body previously noted "Parent Legendary: TBD". The correct parent is #8134. **Dependency direction**: This Epic (#8084) BLOCKS Legendary #8134. Legendary #8134 DEPENDS ON this Epic. --- **Automated by CleverAgents Bot** Supervisor: Epic Planning | Agent: epic-planning-pool-supervisor
Author
Owner

🧩 Child Implementation Issues Created

The following child issues have been created for this Epic and are now tracked under milestone v3.6.0:

# Issue Priority Branch
1 #8203 — feat: implement semantic chunking context strategy for ACMS advanced context assembly 🟠 High feat/acms-semantic-chunking-context-strategy
2 #8204 — feat: integrate Anthropic Claude and Google Gemini as additional LLM backends 🟠 High feat/llm-backends-anthropic-claude-google-gemini
3 #8205 — feat: implement pluggable scope chain resolution extension API for custom context sources 🟠 High feat/pluggable-scope-chain-resolution-extension-api

All 3 child issues block this Epic and must be completed before this Epic can be closed.


Automated by CleverAgents Bot
Supervisor: Epic Planning | Agent: epic-planning-pool-supervisor

## 🧩 Child Implementation Issues Created The following child issues have been created for this Epic and are now tracked under milestone **v3.6.0**: | # | Issue | Priority | Branch | |---|-------|----------|--------| | 1 | #8203 — feat: implement semantic chunking context strategy for ACMS advanced context assembly | 🟠 High | `feat/acms-semantic-chunking-context-strategy` | | 2 | #8204 — feat: integrate Anthropic Claude and Google Gemini as additional LLM backends | 🟠 High | `feat/llm-backends-anthropic-claude-google-gemini` | | 3 | #8205 — feat: implement pluggable scope chain resolution extension API for custom context sources | 🟠 High | `feat/pluggable-scope-chain-resolution-extension-api` | All 3 child issues block this Epic and must be completed before this Epic can be closed. --- **Automated by CleverAgents Bot** Supervisor: Epic Planning | Agent: epic-planning-pool-supervisor
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Reference
cleveragents/cleveragents-core#8084
No description provided.