telegram-researchlisted
Install: claude install-skill warodan/telegram-research
# Telegram research
Three questions, in this order and never mixed: **which channels and groups talk about this** (discovery),
**what words do they use for it** (query craft), **what was actually said** (reading). Step 2 has no input
until step 1 has produced a corpus to mine, and step 3 finds the wrong half of what exists if step 2 is skipped.
## Finding the scripts
Everything runs through `scripts/tg.py`, beside this file. Resolve its path once, at the start of a run:
```bash
for p in "${CLAUDE_SKILL_DIR}" ~/.claude/skills/telegram-research ~/.agents/skills/telegram-research \
./.claude/skills/telegram-research ./.agents/skills/telegram-research; do
[ -n "$p" ] && [ -f "$p/scripts/tg.py" ] && { TG="$p/scripts/tg.py"; break; }
done
if [ -z "$TG" ]; then
echo "telegram-research: no scripts/tg.py in \$CLAUDE_SKILL_DIR, nor in .claude/skills/telegram-research" >&2
echo "or .agents/skills/telegram-research under ~ or ./ - set CLAUDE_SKILL_DIR to the folder holding" >&2
echo "SKILL.md, or run from the project root the skill is installed in." >&2
exit 7
fi
if command -v python3 >/dev/null 2>&1; then python3 "$TG" selftest; else python "$TG" selftest; fi
```
**The refusal is the load-bearing half.** Without it an unmatched loop leaves `$TG` empty, the interpreter is
handed no file, and the shell returns **1** — the one code this skill never produces deliberately, so the
failure reads as "the skill is broken" when it means "it was run from somewhere it cannot