coulomb/markitect-main
Archived
2
0
Fork 0
Code Issues 135 Pull Requests Actions Projects 12 Releases 1 Wiki Activity

feat(llm): add Gemini adapter and process book-1-chapter-05

Browse Source 
Add GeminiAdapter calling Google's Generative Language REST API
(default model: gemini-2.5-flash). Register "gemini" as third
provider in the factory and CLI. Add rate-limit retry with
exponential backoff to the pipeline's _call_llm helper. Increase
default max_tokens from 2000 to 4096.

Process book-1-chapter-05 via Gemini free tier — 1 new entity
extracted (necessaries-conveniencies-and-amusements-of-life),
41 existing entities correctly skipped by dedup. Canonical set
now at 42 unique entities.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
Bernd Worsch
2026-02-11 22:54:37 +01:00
parent 2d1282a61e
commit 880c1d1374
22 changed files with 12008 additions and 57 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
examples/infospace-with-history/INFRA-TASKS.md
View File

@@ -1,10 +1,10 @@
# Markitect Infrastructure Tasks
Issues discovered while building the infospace-with-history example.
These should be addressed in the markitect infrastructure, then the
experiment re-run.
All three have been fixed in commit `706981c` and the pipeline script
refactored to use the fixed infrastructure directly.
## 1. Artifact Repository does not store content
## 1. Artifact Repository does not store content — RESOLVED
**File:** `markitect/prompts/resolver/resolver.py`, line 147-148
**Issue:** `content = f"[Content of {artifact.name} from {space_id}]"` — the
@@ -13,10 +13,11 @@ the SQLiteArtifactRepository stores metadata (digest, name, type) but not
the content itself.
**Impact:** Consumers must maintain their own content cache alongside the
repository, defeating the purpose of centralised artifact storage.
**Fix:** Add a `content` column to the artifacts table, or use a separate
content-addressable store keyed by digest.
**Fix applied:** Added `content` field to `Artifact` model, `content TEXT`
column to SQLite schema (with migration for existing DBs), and replaced
the resolver placeholder with `artifact.content`.
## 2. ContentMacro raw_text defaults to empty string
## 2. ContentMacro raw_text defaults to empty string — RESOLVED
**File:** `markitect/prompts/templates/models.py`, line 46
**Issue:** `raw_text: str = ""` — when macros are constructed programmatically
@@ -24,15 +25,15 @@ content-addressable store keyed by digest.
ContextCompiler then calls `str.replace("", resolved.content)` which inserts
content between every character, producing multi-gigabyte output.
**Impact:** Silent data corruption; compiled prompts become unusable.
**Fix:** Either require `raw_text` in the constructor, or derive it
automatically from `target` (e.g., `raw_text = f"@{{{target}}}"` if not
provided).
**Fix applied:** Added `__post_init__` to `ContentMacro` that auto-derives
`raw_text = f"@{{{self.target}}}"` when not provided.
## 3. No TemplateAnalyzer support for @{target} syntax
## 3. No TemplateAnalyzer support for @{target} syntax — RESOLVED
**File:** `markitect/prompts/templates/analyzer.py`
**Issue:** The TemplateAnalyzer parses `{{kind:target}}` syntax but the
**File:** `markitect/prompts/templates/parser.py`
**Issue:** The MacroParser parses `{{kind:target}}` syntax but the
templates in this example use the simplified `@{target}` syntax. There's
no automatic parsing for this format, requiring manual macro construction.
**Fix:** Add `@{target}` as a recognised shorthand syntax in the analyzer,
auto-detecting it as `{{require:target}}`.
**Fix applied:** Added `SHORTHAND_PATTERN` to `MacroParser` that recognises
`@{target}` and maps it to `MacroKind.REQUIRED`. Updated `has_macros()`,
`count_macros()`, and `find_macro_positions()` accordingly.