Files
aish/config.lua
T
marfrit f26cbd9a3a phase2 amend: __ separator (Bedrock-safe) + post_sse error diagnostics
Phase 7 verify finding from TC #26 against :model cloud:
  HTTP 400 from openrouter→Amazon Bedrock:
  "tools.0.custom.name: String should match pattern
   '^[a-zA-Z0-9_-]{1,128}$'"

Anthropic via Bedrock validates tool names against that regex and
rejects dots. PHASE2 originally chose "." as the namespace separator
("boltzmann.list_dir"); OpenAI tolerated it, Bedrock does not.

Separator switched to "__" (two underscores) everywhere — internal
API matches on-wire shape, no transformation layer:

  - repl.lua:
    - tools_schema builds "alias__name"
    - dispatch_tool_call splits via "^(.-)__(.+)$" (non-greedy → leftmost __)
    - :mcp tool parser uses same split
    - :mcp tools formatter prints "alias__name"
    - HELP block shows <alias__name>
  - safety.lua confirm_tool_call: alias.* glob → alias__* glob
  - config.lua example block: keys rewritten
  - docs/PHASE2.md: amendment header added; §1, §2 row, §3 config.lua
    row, §5 wire-shape JSON examples, §6 auto_approve schema, §7
    meta-cmd table, §12 plan all updated. Original "." references
    preserved in commit history.

Constraint: aliases must not themselves contain "__" so the parse
stays unambiguous. Tool names from MCP servers may have underscores
freely.

Second fix bundled — uninformative broker error:
  Previously "broker error: transport: HTTP response code said error"
  Now      "broker error: transport: HTTP 400: {full body snippet}"

ffi/curl.lua M.post_sse changes:
  - FAILONERROR no longer set (was hiding the response body).
  - raw_body accumulator added alongside the SSE buffer; captures
    every byte regardless of SSE shape.
  - After perform, check status_code via curl_easy_getinfo. On >=400,
    return (nil, "HTTP <code>: <body[:400]>"). 2xx unchanged.
  - End-of-stream SSE flush only runs on 2xx (no false event on
    error bodies that aren't SSE-shaped).
  - Phase 1 callers reading just first return slot stay correct.

End-to-end verified:
  - :model cloud + tools=[boltzmann__read_file ...] +
    "Use boltzmann__read_file with path=/etc/hostname" →
    Claude emits tool_call with name="boltzmann__read_file",
    args='{"path": "/etc/hostname"}'. ok=true, transport clean.
  - Force-bad tool name "bad.name.with.dots" → err string carries
    the full bedrock 400 with the regex-pattern message visible.

TC #26 (sub-loop end-to-end) is now testable against cloud — the
error that blocked it is resolved.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-12 20:04:57 +00:00

96 lines
3.7 KiB
Lua

-- config.lua — model registry, routing rules, user preferences.
-- Loaded with dofile() at startup; returns a plain Lua table.
-- See docs/PHASE0.md §10 for resolution order and full schema.
--
-- Per issue #12: hossenfelder is the canonical single-URL broker. It does
-- model-aware routing server-side (local models on boltzmann; cloud routes
-- through OpenRouter using its own bearer auth — no client-side key here).
-- Discovery: GET http://hossenfelder.fritz.box:8082/v1/models.
local HOSSENFELDER = "http://hossenfelder.fritz.box:8082"
return {
default_model = "fast",
models = {
fast = {
endpoint = HOSSENFELDER,
model = "qwen2.5-coder-1.5b-q4_k_m.gguf",
temperature = 0.2,
},
deep = {
endpoint = HOSSENFELDER,
model = "qwen3-30b-a3b-instruct",
timeout_ms = 1800000, -- 10 min; Nemo on RK3588 is patient work
temperature = 0.1,
},
cloud = {
endpoint = HOSSENFELDER,
model = "anthropic/claude-haiku-4.5",
temperature = 0.2,
},
},
shell = {
known_commands = {
"ls", "cat", "cd", "grep", "find", "cp", "mv", "rm",
"mkdir", "rmdir", "git", "make", "cmake", "gcc", "clang",
"python3", "luajit", "ssh", "scp", "curl", "wget",
},
capture_output = true, -- inject exec output into context
confirm_cmd = true, -- prompt before executing CMD: suggestions
},
context = {
max_turns = 40,
token_budget = 4096,
},
history = {
dir = (os.getenv("HOME") or ".") .. "/.local/share/aish",
},
-- Phase 2 (docs/PHASE2.md): MCP server registry + tool-call policy.
-- The block is OFF by default — connect-at-startup happens only when
-- `servers` is non-empty. Uncomment + adjust per your fleet.
--
-- mcp = {
-- servers = {
-- -- Each entry: alias = { url = "...", auth_token = "..." | auth_env = "..." }
-- -- auth_token literal > auth_env env-var indirection > nil (no auth).
-- -- Aliases become the namespace prefix on tool names sent to the model
-- -- ("<alias>__<tool>" — e.g. "boltzmann__list_dir"). The separator is
-- -- "__" (two underscores) because Anthropic via Bedrock validates tool
-- -- names against ^[a-zA-Z0-9_-]{1,128}$ — dots are rejected.
-- -- Aliases themselves must not contain "__".
-- boltzmann = {
-- url = "http://boltzmann.fritz.box:8080/mcp",
-- auth_env = "BOLTZMANN_MCP_TOKEN",
-- },
-- hertz = {
-- url = "http://hertz.fritz.box:8080/mcp",
-- auth_env = "HERTZ_MCP_TOKEN",
-- },
-- broglie = {
-- url = "http://broglie.fritz.box:8080/mcp", -- LAN-only, no auth
-- },
-- },
--
-- -- Per-call confirm gate auto-approve policy.
-- -- Key forms:
-- -- "<alias>__<tool>" — auto-approve one specific tool
-- -- "<alias>__*" — auto-approve every tool on that server
-- -- Anything not matched falls back to the [y/N] prompt.
-- auto_approve = {
-- ["boltzmann__read_file"] = true,
-- ["boltzmann__list_dir"] = true,
-- ["boltzmann__search_files"] = true,
-- ["hertz__*"] = true, -- trust the hub fully
-- },
--
-- -- Tool-call sub-loop budget per ask_ai turn. Hitting the cap surfaces
-- -- a status and breaks; default 8 if absent.
-- max_tool_depth = 8,
-- },
}