Commit graph

12 commits

Author SHA1 Message Date
Krrish Dholakia
8a696bfc5d fix(headroom-proxy): listen on port 10000 for Render Docker web service 2026-06-29 19:07:02 -07:00
Krrish Dholakia
69755117d4 feat(headroom-proxy): replace with minimal mock server (no ML/Rust, just CCR compress+retrieve) 2026-06-29 19:02:44 -07:00
Krrish Dholakia
48e6fb9387 fix(headroom-proxy): simplify dockerfile, add import smoke test to catch startup failures in build 2026-06-29 18:41:04 -07:00
Krrish Dholakia
93deacc724 fix(headroom-proxy): run sha256sum check from /tmp to match sidecar path format 2026-06-29 18:31:14 -07:00
Krrish Dholakia
9a394427e7 fix(headroom-proxy): use python:3.12 builder with pinned verified rustup, copy site-packages to final stage 2026-06-29 18:29:50 -07:00
Krrish Dholakia
3c7daf6e6c fix(headroom-proxy): use multi-stage build with Rust 1.87 to compile headroom._core extension 2026-06-29 18:25:04 -07:00
Krrish Dholakia
654f610693 fix(headroom-proxy): disable Rust core requirement for Python-only mode on Render 2026-06-29 18:17:32 -07:00
Krrish Dholakia
1fc28518c4 fix(headroom-proxy): revert to port 8787 to match Render service config, add --no-optimize 2026-06-29 18:12:05 -07:00
Krrish Dholakia
d370ed2cc0 fix(headroom-proxy): skip ML model download at startup with --no-optimize 2026-06-29 18:05:47 -07:00
Krrish Dholakia
84b882e201 fix(headroom-proxy): use port 10000 for Render compatibility 2026-06-29 17:42:58 -07:00
Krrish Dholakia
d50061a1e5 feat(prompt-compression): add headroom raw proxy deployment for Render 2026-06-25 22:22:29 -07:00
Krrish Dholakia
65b8660c80 feat(prompt-compression): add headroom compression plugin
Microservice that wraps headroom-ai as a LiteLLM pre_call guardrail.
Compresses tool outputs and JSON arrays before they reach the LLM
provider using headroom's smart_crusher (JSON dedup + schema compression)
and kompress (prose/log compression). Returns compressed structured_messages
via the generic guardrail API, which required adding structured_messages
support to GenericGuardrailAPIResponse.
2026-06-25 19:49:26 -07:00